1 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
3 * cfghooks.c (verify_flow_info): Check that edge probabilities are
6 2017-10-13 Nathan Sidwell <nathan@acm.org>
8 * tree-core.h (tree_contains_struct): Make bool.
9 * tree.c (tree_contains_struct): Likewise.
10 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
11 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
12 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
13 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
14 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
16 2017-10-13 Richard Biener <rguenther@suse.de>
18 * graphite-isl-ast-to-gimple.c
19 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
20 parameters and dominance check.
21 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
22 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
23 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
24 Do not update SSA form here or do intermediate IL verification.
25 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
26 (graphite_initialize): Remove check on the number of loops in
27 the function and inline into graphite_transform_loops.
28 (graphite_finalize): Inline into graphite_transform_loops.
29 (graphite_transform_loops): Perform SSA update and IL verification
31 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
33 2017-10-13 Richard Biener <rguenther@suse.de>
35 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
36 graphite_expression_type_precision): Avoid global constructor
38 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
39 (translate_isl_ast_to_gimple::graphite_expr_type): Add type
41 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
42 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
43 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
44 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
46 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
49 * config/i386/i386.h (ix86_red_zone_size): New.
50 * config/i386/i386.md (push peephole2s): Replace
51 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
53 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
54 Alan Hayward <alan.hayward@arm.com>
55 David Sherwood <david.sherwood@arm.com>
57 * combine.c (can_change_dest_mode): Reject changes in
60 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
61 Alan Hayward <alan.hayward@arm.com>
62 David Sherwood <david.sherwood@arm.com>
64 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
65 (expand_debug_source_expr): Likewise.
66 * combine.c (combine_simplify_rtx): Likewise.
67 * cse.c (fold_rtx): Likewise.
68 * fwprop.c (canonicalize_address): Likewise.
69 * targhooks.c (default_shift_truncation_mask): Likewise.
71 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
72 Alan Hayward <alan.hayward@arm.com>
73 David Sherwood <david.sherwood@arm.com>
75 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
76 (widened_mode): Likewise.
77 (expand_unop): Likewise.
78 * ree.c (transform_ifelse): Likewise.
79 (merge_def_and_ext): Likewise.
80 (combine_reaching_defs): Likewise.
81 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
83 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
84 Alan Hayward <alan.hayward@arm.com>
85 David Sherwood <david.sherwood@arm.com>
87 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
88 * combine.c (gen_lowpart_for_combine): Likewise.
89 * dwarf2out.c (rtl_for_decl_location): Likewise.
90 * final.c (alter_subreg): Likewise.
91 * rtlhooks.c (gen_lowpart_general): Likewise.
92 (gen_lowpart_if_possible): Likewise.
94 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
95 Alan Hayward <alan.hayward@arm.com>
96 David Sherwood <david.sherwood@arm.com>
98 * calls.c (expand_call): Use subreg_lowpart_offset.
99 * cse.c (cse_insn): Likewise.
100 * regcprop.c (copy_value): Likewise.
101 (copyprop_hardreg_forward_1): Likewise.
103 2017-10-13 Jakub Jelinek <jakub@redhat.com>
106 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
107 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
108 =Q constraints to +Q and into insn condition add check
109 that operands[0] and operands[1] are equal.
110 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
111 =Q constraints to +Q and into insn condition add check
112 that operands[0] is equal to either operands[1] or operands[2].
115 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
116 instead of handling MINUS_EXPR twice (once for each argument),
117 canonicalize operand order and handle just once, use rtype where
118 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
121 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
122 any values of __C while still being pattern recognizable as a simple
125 2017-10-13 Richard Biener <rguenther@suse.de>
127 PR tree-optimization/82451
129 2017-10-02 Richard Biener <rguenther@suse.de>
131 PR tree-optimization/82355
132 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
133 a mapping for the enclosing loop but avoid generating one for
135 (copy_bb_and_scalar_dependences): Remove premature codegen
136 error on PHIs in blocks duplicated into multiple places.
137 * graphite-scop-detection.c
138 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
139 in the region use it as loop and nest to analyze the DR in.
140 (try_generate_gimple_bb): Likewise.
141 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
142 (add_loop_constraints): For blocks in a loop not in the region
143 create a dimension with a single iteration.
144 * sese.h (gbb_loop_at_index): Remove assert.
146 * cfgloop.c (loop_preheader_edge): For the loop tree root
147 return the single successor of the entry block.
148 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
149 Reset the SCEV hashtable and niters.
150 * graphite-scop-detection.c
151 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
152 assert that we only have POLYNOMIAL_CHREC that vary in loops
153 contained in the region.
154 (scop_detection::graphite_can_represent_expr): Adjust.
155 (scop_detection::stmt_has_simple_data_refs_p): For loops
156 not in the region set loop to NULL. The nest is now the
157 entry edge to the region.
158 (try_generate_gimple_bb): Likewise.
159 * sese.c (scalar_evolution_in_region): Adjust for
160 instantiate_scev change.
161 * tree-data-ref.h (graphite_find_data_references_in_stmt):
162 Make nest parameter the edge into the region.
163 (create_data_ref): Likewise.
164 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
165 entry edge into a region and adjust instantiate_scev calls.
166 (create_data_ref): Likewise.
167 (graphite_find_data_references_in_stmt): Likewise.
168 (find_data_references_in_stmt): Pass the loop preheader edge
169 from the nest argument.
170 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
171 parameter the edge into the region.
172 (instantiate_parameters): Use the loop preheader edge as entry.
173 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
175 (get_instantiated_value_entry): Make instantiate_below parameter
176 the edge into the region.
177 (instantiate_scev_name): Likewise. Adjust dominance checks,
178 when we cannot use loop-based instantiation instantiate by
179 walking use-def chains.
180 (instantiate_scev_poly): Adjust.
181 (instantiate_scev_binary): Likewise.
182 (instantiate_scev_convert): Likewise.
183 (instantiate_scev_not): Likewise.
184 (instantiate_array_ref): Remove.
185 (instantiate_scev_3): Likewise.
186 (instantiate_scev_2): Likewise.
187 (instantiate_scev_1): Likewise.
188 (instantiate_scev_r): Do not blindly handle N-operand trees.
189 Do not instantiate array-refs. Handle all constants and invariants.
190 (instantiate_scev): Make instantiate_below parameter
191 the edge into the region.
192 (resolve_mixers): Use the loop preheader edge for the region
193 parameter to instantiate_scev_r.
194 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
196 2017-10-13 Richard Biener <rguenther@suse.de>
198 PR tree-optimization/82525
199 * graphite-isl-ast-to-gimple.c
200 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
202 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
203 Fail code generation when we cannot represent the isl integer.
204 (binary_op_to_tree): Elide modulo operations that are no-ops
205 in the type we code generate. Remove now superfluous code
208 2017-10-13 Richard Biener <rguenther@suse.de>
210 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
211 (scop_detection::harmful_loop_in_region): Remove premature
213 (scop_detection::graphite_can_represent_scev): We can handle
214 pointer IVs just fine.
216 2017-10-13 Alan Modra <amodra@gmail.com>
218 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
219 "Clobbers and Scratch Registers". Add paragraph on
220 alternative to clobbers for scratch registers and OpenBLAS
223 2017-10-13 Alan Modra <amodra@gmail.com>
225 * doc/extend.texi (Clobbers): Correct vax example. Delete old
226 example of a memory input for a string of known length. Move
227 commentary out of table. Add a number of new examples
228 covering array memory inputs.
230 2017-10-12 Martin Liska <mliska@suse.cz>
232 PR tree-optimization/82493
233 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
234 (test_range_functions): New function.
235 (sbitmap_c_tests): Likewise.
236 * selftest-run-tests.c (selftest::run_tests): Run new tests.
237 * selftest.h (sbitmap_c_tests): New function.
239 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
241 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
243 * config/rs6000/amo.h: Fix spacing issue.
245 2017-10-12 Jakub Jelinek <jakub@redhat.com>
248 * config/i386/i386.md (*ashl<mode>3_mask_1,
249 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
250 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
253 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
255 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
256 (profile_probability): Set max_probability
257 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
259 * profile-count.c (profile_probability::differs_from_p): Do not
260 rely on max_probaiblity == 10000
262 2017-10-12 Jeff Law <law@redhat.com>
264 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
267 2017-10-12 Martin Sebor <msebor@redhat.com>
271 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
272 (handle_alias_pairs): Call it.
273 * common.opt (-Wattribute-alias): New option.
274 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
275 * doc/invoke.texi (-Wattribute-alias): Document.
277 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
280 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
282 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
284 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
285 (make_hard_regno_born, make_hard_regno_dead): Update
286 bb_killed_pseudos and bb_gen_pseudos.
288 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
290 * x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
292 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
294 * config/alpha/alpha.c (alpha_split_conditional_move):
295 Use std::swap instead of manually swapping.
296 (alpha_stdarg_optimize_hook): Ditto.
297 (alpha_canonicalize_comparison): Ditto.
299 2017-10-12 Bin Cheng <bin.cheng@arm.com>
301 * tree-loop-distribution.c (struct builtin_info): New struct.
302 (struct partition): Refactor fields into struct builtin_info.
303 (partition_free): Free struct builtin_info.
304 (build_size_arg_loc, build_addr_arg_loc): Delete.
305 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
306 information from struct builtin_info.
307 (find_single_drs): New function refactored from classify_partition.
308 Also moved builtin validity checks to this function.
309 (compute_access_range, alloc_builtin): New functions.
310 (classify_builtin_st, classify_builtin_ldst): New functions.
311 (classify_partition): Refactor code into functions find_single_drs,
312 classify_builtin_st and classify_builtin_ldst.
313 (distribute_loop): Don't do runtime alias check when distributing
315 (find_seed_stmts_for_distribution): New function.
316 (pass_loop_distribution::execute): Refactor code finding seed
317 stmts into above function. Support distribution for the innermost
318 two-level loop nest. Adjust dump information.
320 2017-10-12 Bin Cheng <bin.cheng@arm.com>
322 * tree-loop-distribution.c: Adjust the general comment.
323 (NUM_PARTITION_THRESHOLD): New macro.
324 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
325 (classify_partition): Skip builtin pattern of loop nest's inner loop.
326 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
327 in call to build_partition_graph.
328 (finalize_partitions): New parameter. Make loop distribution more
329 conservative by fusing more partitions.
330 (distribute_loop): Don't do runtime alias check in case of loop nest
332 (find_seed_stmts_for_distribution): New function.
333 (prepare_perfect_loop_nest): New function.
334 (pass_loop_distribution::execute): Refactor code finding seed stmts
335 and loop nest into above functions. Support loop nest distribution.
336 Adjust dump information accordingly.
338 2017-10-12 Bin Cheng <bin.cheng@arm.com>
340 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
341 and set PTYPE_SEQUENTIAL for merged partition.
343 2017-10-12 Richard Biener <rguenther@suse.de>
345 PR tree-optimization/69728
347 2017-09-19 Richard Biener <rguenther@suse.de>
349 PR tree-optimization/69728
350 * graphite-sese-to-poly.c (schedule_error): New global.
351 (add_loop_schedule): Handle empty domain by failing the
353 (build_original_schedule): Handle schedule_error.
355 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
356 domain by returning an unchanged schedule.
358 2017-10-12 Jakub Jelinek <jakub@redhat.com>
360 * genrecog.c (validate_pattern): For VEC_SELECT verify that
361 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
363 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
365 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
368 2017-10-12 Jakub Jelinek <jakub@redhat.com>
371 * expr.c (store_field): Don't optimize away bitsize == 0 store
372 from CALL_EXPR with addressable return type.
374 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
376 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
377 * config/rs6000/rs6000.md (sel): Delete mode attribute.
378 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
379 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
380 TARGET_ISEL instead of TARGET_ISEL<sel>.
382 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
384 * config/rs6000/rs6000.c
385 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
387 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
389 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
391 (reg_or_cint_operand): Fix comment.
392 (reg_or_zero_operand): New predicate.
393 * config/rs6000/rs6000-protos.h (output_isel): Delete.
394 * config/rs6000/rs6000.c (output_isel): Delete.
395 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
396 instead of reg_or_cint_operand. Output instruction directly (not via
398 (isel_unsigned_<mode>): Ditto.
399 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
400 gpc_reg_operand. Add an instruction alternative for this. Output
401 instruction directly.
402 (*isel_reversed_unsigned_<mode>): Ditto.
404 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
406 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
407 (TARGET_CANONICALIZE_COMPARISON): Define.
409 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
412 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
413 Check whether the dest is REG before adding REG_EQUIV note.
415 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
418 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
420 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
421 (make_hard_regno_born, make_hard_regno_dead): Update
422 bb_killed_pseudos and bb_gen_pseudos.
424 2017-10-11 Nathan Sidwell <nathan@acm.org>
426 * incpath.h (enum incpath_kind): Name enum, prefix values.
427 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
428 * incpath.c (heads, tails): Use INC_MAX.
429 (add_env_var_paths, add_standard_paths): Use incpath_kind.
430 (merge_include_chains, split_quote_chain,
431 register_include_chains): Update incpath_kind names.
432 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
433 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
435 (add_framework_path, darwin_register_objc_includes): Likewise.
436 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
438 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
440 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
441 Do not use float_operator operator predicate.
442 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
443 * config/i386/predicates.md (float_operator): Remove predicate.
445 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
447 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
448 (*jccxf_i387): Ditto.
449 (*jcc<mode>_i387): Ditto.
450 (*jccu<mode>_i387): Ditto.
451 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
452 (*jcc_*_i387 splitters): Remove.
453 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
454 * config/i386/i386.c (ix86_split_fp_branch): Remove.
455 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
458 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
460 * profile-count.h (slow_safe_scale_64bit): New function.
461 (safe_scale_64bit): New inline.
462 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
463 * profile-count.c: Include wide-int.h
464 (slow_safe_scale_64bit): New.
466 2017-10-11 Nathan Sidwell <nathan@acm.org>
468 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
469 HAS_DECL_ASSEMBLER_NAME_P.
470 * gimple-expr.c (gimple_decl_printable_name: Check
471 HAS_DECL_ASSEMBLER_NAME_P too.
472 * ipa-utils.h (type_in_anonymous_namespace_p): Check
473 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
474 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
475 * passes.c (rest_of_decl_compilation): Check
476 HAS_DECL_ASSEMBLER_NAME_P too.
477 * recog.c (verify_changes): Likewise.
478 * tree-pretty-print.c (dump_decl_name): Likewise.
479 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
481 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
482 (SET_DECL_ASSEMBLER_NAME): Use it.
483 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
484 (COPY_DECL_ASSEMBLER_NAME): Likewise.
485 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
487 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
489 * config.gcc (i386, x86_64): Add extra objects.
490 * i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
491 (ix86_min_insn_size): Declare.
492 (ix86_issue_rate): Declare.
493 (ix86_adjust_cost): Declare.
494 (ia32_multipass_dfa_lookahead): Declare.
495 (ix86_macro_fusion_p): Declare.
496 (ix86_macro_fusion_pair_p): Declare.
497 (ix86_bd_has_dispatch): Declare.
498 (ix86_bd_do_dispatch): Declare.
499 (ix86_core2i7_init_hooks): Declare.
500 (ix86_atom_sched_reorder): Declare.
501 * i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
502 (COSTS_N_BYTES): Move to x86-tune-costs.h.
503 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
504 (rip_relative_addr_p): Rename to ...
505 (ix86_rip_relative_addr_p): ... this one; export.
506 (memory_address_length): Update.
507 (ix86_issue_rate): Move to x86-tune-sched.c.
508 (ix86_flags_dependent): Move to x86-tune-sched.c.
509 (ix86_agi_dependent): Move to x86-tune-sched.c.
510 (exact_dependency_1): Move to x86-tune-sched.c.
511 (exact_store_load_dependency): Move to x86-tune-sched.c.
512 (ix86_adjust_cost): Move to x86-tune-sched.c.
513 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
514 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
515 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
516 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
517 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
518 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
519 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
520 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
521 (min_insn_size): Rename to ...
522 (ix86_min_insn_size): ... this one; export.
523 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
524 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
525 (core2i7_first_cycle_multipass_backtrack): Move to
526 x86-tune-sched-core.c.
527 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
528 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
529 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
530 (ix86_avoid_jump_mispredicts): Update.
531 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
532 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
533 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
534 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
535 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
536 (MAX_INSN): Move to ix86-tune-sched-bd.c.
537 (MAX_IMM): Move to ix86-tune-sched-bd.c.
538 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
539 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
540 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
541 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
542 (MAX_STORE): Move to ix86-tune-sched-bd.c.
543 (BIG): Move to ix86-tune-sched-bd.c.
544 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
545 (enum insn_path): Move to ix86-tune-sched-bd.c.
546 (get_mem_group): Move to ix86-tune-sched-bd.c.
547 (is_cmp): Move to ix86-tune-sched-bd.c.
548 (dispatch_violation): Move to ix86-tune-sched-bd.c.
549 (is_branch): Move to ix86-tune-sched-bd.c.
550 (is_prefetch): Move to ix86-tune-sched-bd.c.
551 (init_window): Move to ix86-tune-sched-bd.c.
552 (allocate_window): Move to ix86-tune-sched-bd.c.
553 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
554 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
555 (process_end_window): Move to ix86-tune-sched-bd.c.
556 (allocate_next_window): Move to ix86-tune-sched-bd.c.
557 (find_constant): Move to ix86-tune-sched-bd.c.
558 (get_num_immediates): Move to ix86-tune-sched-bd.c.
559 (has_immediate): Move to ix86-tune-sched-bd.c.
560 (get_insn_path): Move to ix86-tune-sched-bd.c.
561 (get_insn_group): Move to ix86-tune-sched-bd.c.
562 (count_num_restricted): Move to ix86-tune-sched-bd.c.
563 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
564 (add_insn_window): Move to ix86-tune-sched-bd.c.
565 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
566 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
567 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
568 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
569 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
570 (do_dispatch): Move to ix86-tune-sched-bd.c.
571 (has_dispatch): Move to ix86-tune-sched-bd.c.
572 * i386/t-i386: Add new object files.
573 * i386/x86-tune-costs.h: New file.
574 * i386/x86-tune-sched-atom.c: New file.
575 * i386/x86-tune-sched-bd.c: New file.
576 * i386/x86-tune-sched-core.c: New file.
577 * i386/x86-tune-sched.c: New file.
579 2017-10-11 Liu Hao <lh_mouse@126.com>
581 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
582 the generic version below instead.
583 (should_colorize): Recognize Windows consoles as terminals
585 * pretty-print.c [__MINGW32__] (write_all): New function.
586 [__MINGW32__] (find_esc_head): Likewise.
587 [__MINGW32__] (find_esc_terminator): Likewise.
588 [__MINGW32__] (eat_esc_sequence): Likewise.
589 [__MINGW32__] (mingw_ansi_fputs): New function that handles
591 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
594 2017-10-11 Richard Biener <rguenther@suse.de>
596 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
597 Properly call analyze_scalar_evolution with the loop of the stmt.
599 2017-10-11 Richard Biener <rguenther@suse.de>
601 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
602 * tree-core.h (tree_base): Add chrec_var union member.
603 * tree.h (CHREC_VAR): Remove.
604 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
605 * tree-chrec.h (build_polynomial_chrec): Adjust.
606 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
607 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
609 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
611 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
612 * match.pd: ... here.
613 ((T) X == (T) Y): Relax condition.
615 2017-10-11 Bin Cheng <bin.cheng@arm.com>
617 PR tree-optimization/82472
618 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
620 (break_alias_scc_partitions): Update postorder number.
622 2017-10-11 Martin Liska <mliska@suse.cz>
625 * opts.c (parse_no_sanitize_attribute): Do not use error_value
627 * opts.h (parse_no_sanitize_attribute): Remove last argument.
629 2017-10-11 Martin Liska <mliska@suse.cz>
631 * print-rtl.c (print_insn): Move declaration of idbuf
632 to same scope as name.
634 2017-10-11 Martin Liska <mliska@suse.cz>
639 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
642 2017-10-11 Martin Liska <mliska@suse.cz>
645 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
648 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
650 * config.gcc (default_gnu_indirect_function): Default to yes for
651 arm*-*-linux* with glibc.
653 2017-10-11 Richard Biener <rguenther@suse.de>
655 * tree-scalar-evolution.c (get_scalar_evolution): Handle
656 default-defs and types we do not want to analyze.
657 (interpret_loop_phi): Replace unreachable code with an assert.
658 (compute_scalar_evolution_in_loop): Remove and inline ...
659 (analyze_scalar_evolution_1): ... here, replacing condition with
660 what makes the intent clearer. Remove handling of cases
661 get_scalar_evolution now handles.
663 2017-10-10 Jim Wilson <wilson@tuliptree.org>
665 PR rtl-optimization/81434
666 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
667 comment for main loop. In sched_group_found if, also add checks for
668 pass and min_cost_group.
670 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
672 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
673 (rs6000_insn_cost): New function.
674 * config/rs6000/rs6000.md (cost): New attribute.
676 2017-10-10 Jakub Jelinek <jakub@redhat.com>
677 H.J. Lu <hongjiu.lu@intel.com>
681 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
682 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
683 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
684 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
685 to OPTION_MASK_ISA_AVX512VL - builtins that have both
686 OPTION_MASK_ISA_MMX and some other bit set require both
687 mmx and the ISAs without the mmx bit.
688 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
689 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
690 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
691 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
692 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
693 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
694 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
695 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
696 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
697 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
698 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
699 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
700 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
701 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
702 Add OPTION_MASK_ISA_MMX.
704 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
706 * config.gcc (armv7*-*-freebsd*): New target.
707 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
709 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
711 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
712 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
715 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
717 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
718 a template parameter.
719 (WIDE_INT_REF_FOR): Update accordingly.
720 * tree.h (wi::int_traits <const_tree>): Delete.
721 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
722 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
723 (wi::tree_to_wide_ref): New typedef.
724 (wi::to_wide): New function.
725 * calls.c (get_size_range): Use wi::to_wide when operating on
727 * cgraph.c (cgraph_node::create_thunk): Likewise.
728 * config/i386/i386.c (ix86_data_alignment): Likewise.
729 (ix86_local_alignment): Likewise.
730 * dbxout.c (stabstr_O): Likewise.
731 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
732 * expr.c (const_vector_from_tree): Likewise.
733 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
734 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
735 (fold_negate_expr_1, int_const_binop_1, const_binop)
736 (fold_convert_const_int_from_real, optimize_bit_field_compare)
737 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
738 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
739 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
740 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
741 (fold_not_const, round_up_loc): Likewise.
742 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
743 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
744 (alloca_call_type): Likewise.
745 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
746 * godump.c (go_output_typedef): Likewise.
747 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
748 * internal-fn.c (get_min_precision): Likewise.
749 * ipa-cp.c (ipcp_store_vr_results): Likewise.
750 * ipa-polymorphic-call.c
751 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
752 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
753 (ipa_modify_call_arguments): Likewise.
754 * match.pd: Likewise.
755 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
756 * print-tree.c (print_node_brief, print_node): Likewise.
757 * stmt.c (expand_case): Likewise.
758 * stor-layout.c (layout_type): Likewise.
759 * tree-affine.c (tree_to_aff_combination): Likewise.
760 * tree-cfg.c (group_case_labels_stmt): Likewise.
761 * tree-data-ref.c (dr_analyze_indices): Likewise.
762 (prune_runtime_alias_test_list): Likewise.
763 * tree-dump.c (dequeue_and_dump): Likewise.
764 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
765 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
766 * tree-pretty-print.c (dump_generic_node): Likewise.
767 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
768 (simple_iv_with_niters): Likewise.
769 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
770 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
771 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
772 * tree-ssa-loop-niter.c (split_to_var_and_offset)
773 (refine_value_range_using_guard, number_of_iterations_ne_max)
774 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
775 (get_cst_init_from_scev, record_nonwrapping_iv)
776 (scev_var_range_cant_overflow): Likewise.
777 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
778 * tree-ssa-pre.c (compute_avail): Likewise.
779 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
780 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
781 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
782 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
783 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
784 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
785 (dump_case_nodes, try_switch_expansion): Likewise.
786 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
787 (vect_do_peeling): Likewise.
788 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
789 * tree-vect-stmts.c (vectorizable_load): Likewise.
790 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
791 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
792 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
793 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
794 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
795 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
796 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
797 (evrp_dom_walker::before_dom_children): Likewise.
798 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
799 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
800 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
801 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
802 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
803 (get_range_pos_neg): Likewise.
804 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
805 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
806 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
807 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
808 * config/avr/avr.c (avr_fold_builtin): Likewise.
809 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
810 * config/msp430/msp430.c (msp430_attr): Likewise.
811 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
812 * config/powerpcspe/powerpcspe-c.c
813 (altivec_resolve_overloaded_builtin): Likewise.
814 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
815 (rs6000_expand_ternop_builtin): Likewise.
816 * config/rs6000/rs6000-c.c
817 (altivec_resolve_overloaded_builtin): Likewise.
818 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
819 (rs6000_expand_ternop_builtin): Likewise.
820 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
822 2017-10-10 Bin Cheng <bin.cheng@arm.com>
824 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
825 when copying loop nest with only one inner loop.
827 2017-10-10 Richard Biener <rguenther@suse.de>
829 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
830 blocks if SCEV is active.
831 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
833 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
834 (scev_initialize): Assert we are not yet initialized.
836 2017-10-10 Bin Cheng <bin.cheng@arm.com>
838 * tree-loop-distribution.c (generate_loops_for_partition): Remove
839 inner loop's exit stmt by making it always exit the loop, otherwise
840 we would generate an infinite empty loop.
842 2017-10-10 Bin Cheng <bin.cheng@arm.com>
844 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
845 renaming variables in new preheader if it's deleted.
847 2017-10-10 Bin Cheng <bin.cheng@arm.com>
849 * tree-loop-distribution.c (struct partition): Remove unused field
850 loops of the structure.
851 (partition_alloc, partition_free): Ditto.
852 (build_rdg_partition_for_vertex): Ditto.
854 2017-10-09 Jeff Law <law@redhat.com>
856 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
857 return type to match prototype and documentation.
859 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
861 * config/rs6000/rs6000.c (processor_costs): Move to ...
862 * config/rs6000/rs6000.h: ... here.
863 (rs6000_cost): Declare.
865 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
867 * except.c (setjmp_fn): New global variable.
868 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
869 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
870 if DONT_USE_BUILTIN_SETJMP is defined.
872 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
874 * target.def (insn_cost): New hook.
875 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
876 * doc/tm.texi: Regenerate.
877 * rtlanal.c (insn_cost): Use the new hook.
879 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
881 * combine.c (combine_validate_cost): Compute the new insn_cost,
882 not just pattern_cost.
883 (try_combine): Adjust comment.
885 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
887 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
889 * combine.c (uid_insn_cost): Adjust comment.
890 (combine_validate_cost): Adjust comment. Use pattern_cost instead
892 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
893 * dse.c (find_shift_sequence): Ditto.
894 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
895 (bb_valid_for_noce_process_p): Use pattern_cost.
896 * rtl.h (insn_rtx_cost): Delete.
897 (pattern_cost): New prototype.
898 (insn_cost): New prototype.
899 * rtlanal.c (insn_rtx_cost): Rename to...
900 (pattern_cost): ... this.
903 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
905 * config/i386/i386.md (*jcc_2): Remove insn pattern.
906 (*jcc<mode>_0_r_i387): Ditto.
907 (*jccxf_r_i387): Ditto.
908 (*jcc<mode>_r_i387): Ditto.
909 (*jccu<mode>_r_i387): Ditto.
910 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
911 (*jcc): Rename from *jcc_1.
913 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
915 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
916 deferred rescans after the lvx/stvx recombination pre-pass.
918 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
920 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
921 memory operation instruction support.
922 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
923 (rs6000-ibm-aix[789]*): Likewise.
924 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
925 Document new functions.
927 2017-10-09 Richard Biener <rguenther@suse.de>
929 PR tree-optimization/82397
930 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
931 equality only for semantically equal trees.
933 2017-10-09 Richard Biener <rguenther@suse.de>
935 PR tree-optimization/82449
936 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
937 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
938 allow constant addresses.
939 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
942 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
944 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
947 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
950 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
953 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
956 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
958 2017-10-09 Jakub Jelinek <jakub@redhat.com>
961 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
962 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
964 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
966 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
967 (WI_BINARY_PREDICATE_RESULT): Likewise.
968 (wi::binary_traits::operator_result): New type.
969 (wi::binary_traits::predicate_result): Likewise.
970 (generic_wide_int::operator~, unary generic_wide_int::operator-)
971 (generic_wide_int::operator==, generic_wide_int::operator!=)
972 (generic_wide_int::operator&, generic_wide_int::and_not)
973 (generic_wide_int::operator|, generic_wide_int::or_not)
974 (generic_wide_int::operator^, generic_wide_int::operator+
975 (binary generic_wide_int::operator-, generic_wide_int::operator*):
977 (operator~, unary operator-, operator==, operator!=, operator&)
978 (operator|, operator^, operator+, binary operator-, operator*): New
980 * expr.c (get_inner_reference): Use wi::bit_and_not.
981 * fold-const.c (fold_binary_loc): Likewise.
982 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
983 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
984 (bit_value_binop): Likewise.
985 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
986 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
987 (extract_range_from_binary_expr_1): Likewise.
988 (masked_increment): Likewise.
989 (simplify_bit_ops_using_ranges): Likewise.
991 2017-10-09 Martin Jambor <mjambor@suse.cz>
994 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
995 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
996 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
998 (hsa_fixup_mov_insn_type): New function.
999 (hsa_op_with_type::get_in_type): Use it.
1000 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
1001 immediates in an assert.
1002 (hsa_op_with_type::extend_int_to_32bit): New method.
1003 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
1004 types. Convert to dest type if necessary.
1005 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
1006 (reg_for_gimple_ssa): Pass false as min32int to
1007 hsa_type_for_scalar_tree_type.
1008 (gen_hsa_addr): Fixup type when creating addresable temporary.
1009 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
1010 (gen_hsa_unary_operation): Extend operands and convert to dest type if
1011 necessary. Call hsa_fixup_mov_insn_type.
1012 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
1013 extend operands and convert to dest type if necessary.
1014 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
1015 to dest type if necessary.
1016 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
1017 if conversion nt necessary and size matches.
1018 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
1019 to dest type if necessary.
1020 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
1021 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
1023 (gen_hsa_clrsb): Likewise.
1024 (gen_hsa_ffs): Likewise.
1025 (gen_hsa_divmod): Extend operands and convert to dest type if
1027 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
1029 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
1031 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
1032 Remove empty default arguments. Use a brace block as output
1034 (conditional return): Ditto.
1036 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
1037 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
1039 (group_ending_nop): Ditto.
1040 (doloop_end): Ditto.
1041 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
1042 (splitters for those): Ditto.
1044 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
1046 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
1047 a conditional jump (and the compare for it) so that pc_rtx is the
1049 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
1050 for the deleted and renamed ctr<mode>_internal[234] patterns.
1051 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
1052 Delete second conditional return pattern.
1053 (ctr<mode>_internal2): Delete this second bdnz pattern.
1054 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
1055 (ctr<mode>_internal4): Delete this second bdz pattern.
1057 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
1059 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
1060 (always_initialized_rtx_for_ssa_name_p): New predicate.
1061 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
1062 (finish_out_of_ssa): Free new field of SA.
1063 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
1064 * tree-ssa-coalesce.c: Include tree-ssa.h.
1065 (get_parm_default_def_partitions): Remove extern keyword.
1066 (get_undefined_value_partitions): New function.
1067 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
1068 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
1070 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
1072 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
1074 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
1076 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
1078 * i386.c (ix86_expand_set_or_movmem): Disable 512bit loops for targets
1079 that preffer 128bit.
1081 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
1083 * i386.c (has_dispatch): Disable for Ryzen.
1085 2017-10-08 Olivier Hainque <hainque@adacore.com>
1087 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
1088 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
1090 (thumb_set_return_address): Likewise.
1092 2017-10-08 Olivier Hainque <hainque@adacore.com>
1094 * common/config/arm/arm-common.c (arm_except_unwind_info):
1095 Handle DWARF2_UNWIND_INFO.
1097 2017-10-07 Michael Collison <michael.collison@arm.com>
1099 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
1102 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
1104 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
1105 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
1106 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
1107 defined, force the creation of a new block for a dispatch label.
1109 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
1111 * invoke.texi (Wsuggest-attribute=cold): Document.
1112 * common.opt (Wsuggest-attribute=cold): New
1113 * ipa-pure-const.c (warn_function_cold): New function.
1114 * predict.c (compute_function_frequency): Use it.
1115 * predict.h (warn_function_cold): Declare.
1117 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
1119 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
1122 2017-10-06 Martin Liska <mliska@suse.cz>
1124 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
1125 keyword for member functions.
1126 (struct sanopt_tree_couple): New struct.
1127 (struct sanopt_tree_couple_hash): New function.
1128 (struct sanopt_ctx): Add new hash_map.
1129 (has_dominating_ubsan_ptr_check): New function.
1130 (record_ubsan_ptr_check_stmt): Likewise.
1131 (maybe_optimize_ubsan_ptr_ifn): Likewise.
1132 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
1133 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
1135 2017-10-06 Sudakshina Das <sudi.das@arm.com>
1138 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
1139 aarch64_simd_valid_immediate on CONST_VECTORs.
1140 (aarch64_reg_or_bic_imm): Likewise.
1142 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
1144 PR rtl-optimization/82396
1145 * haifa-sched.c (ready_sort_real): Disable qsort checking.
1147 2017-10-06 Sebastian Pop <sebpop@gmail.com>
1149 * graphite-dependences.c (scop_get_reads): Move code to...
1150 (scop_get_must_writes): Move code to...
1151 (scop_get_may_writes): Move code to...
1152 (scop_get_reads_and_writes): ... here.
1153 (scop_get_dependences): Call scop_get_reads_and_writes.
1155 2017-10-06 Jakub Jelinek <jakub@redhat.com>
1157 PR tree-optimization/82434
1158 * fold-const.h (can_native_encode_type_p,
1159 can_native_encode_string_p): Remove.
1160 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
1161 don't encode anything, just return what would be otherwise returned.
1162 (native_encode_fixed, native_encode_complex, native_encode_vector):
1164 (native_encode_string): Likewise. Inline by hand
1165 can_native_encode_string_p.
1166 (can_native_encode_type_p): Remove.
1167 (can_native_encode_string_p): Remove.
1168 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
1169 STRING_CSTs using can_native_encode_string_p, test all
1170 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
1171 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
1172 argument from native_encode_expr.
1173 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
1174 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
1177 2017-10-06 Richard Biener <rguenther@suse.de>
1179 PR tree-optimization/82397
1180 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
1181 operand_equal_p but rely on data_ref_compare_tree for detecting
1183 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
1184 to match up with dr_group_sort_cmp.
1186 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1189 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
1191 * config/s390/s390-builtin-types.def: Regenerate.
1193 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1196 * config/s390/s390-builtin-types.def: Regenerate.
1197 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
1198 Change flag from B_VXE to B_VX.
1199 (s390_vec_min_dbl): Remove B_VXE flag.
1201 2017-10-06 Richard Biener <rguenther@suse.de>
1203 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
1204 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
1205 translate_isl_ast_to_gimple::is_valid_rename,
1206 translate_isl_ast_to_gimple::get_rename,
1207 translate_isl_ast_to_gimple::get_def_bb_for_const,
1208 translate_isl_ast_to_gimple::get_new_name,
1209 translate_isl_ast_to_gimple::collect_all_ssa_names,
1210 translate_isl_ast_to_gimple::copy_loop_phi_args,
1211 translate_isl_ast_to_gimple::collect_all_ssa_names,
1212 translate_isl_ast_to_gimple::copy_loop_phi_args,
1213 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
1214 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
1215 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
1216 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
1217 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
1218 translate_isl_ast_to_gimple::copy_cond_phi_args,
1219 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
1220 translate_isl_ast_to_gimple::edge_for_new_close_phis,
1221 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
1222 translate_isl_ast_to_gimple::rename_uses,
1223 translate_isl_ast_to_gimple::rename_all_uses): Remove.
1224 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
1225 (set_rename_for_each_def): Likewise.
1226 (graphite_copy_stmts_from_block): Handle debug stmt resetting
1227 here. Handle rewriting SCEV analyzable uses here.
1228 (copy_bb_and_scalar_dependences): Generate code for PHI
1230 (graphite_regenerate_ast_isl): Adjust.
1231 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
1232 (add_write, add_read): New functions.
1233 (build_cross_bb_scalars_def): Use it and simplify.
1234 (build_cross_bb_scalars_use): Likewise.
1235 (graphite_find_cross_bb_scalar_vars): Inline into...
1236 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
1237 simulating out-of-SSA. Compute liveout and add dependencies.
1238 (build_scops): Force an empty entry block.
1239 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
1241 (sese_build_liveouts): Declare.
1242 (sese_trivially_empty_bb_p): Likewise.
1243 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
1244 compute liveout and debug_liveout.
1245 (sese_bad_liveouts_use): Remove.
1246 (sese_reset_debug_liveouts_bb): Likewise.
1247 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
1248 (sese_build_liveouts): Build liveout and debug_liveout and store
1250 (new_sese_info): Adjust.
1251 (free_sese_info): Likewise.
1252 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
1253 do not build liveout here.
1254 (move_sese_in_condition): Adjust region entry.
1255 (scev_analyzable_p): Match up with chrec_apply requirements.
1256 (sese_trivially_empty_bb_p): New.
1257 * tree-into-ssa.c (get_reaching_def): Properly support generating
1258 default-defs for incremental rewrite of anonymous names.
1260 2017-10-06 Richard Biener <rguenther@suse.de>
1262 * graphite-sese-to-poly.c (extract_affine): For casts increasing
1263 precision do not perform modulo reduction.
1265 2017-10-06 Richard Biener <rguenther@suse.de>
1267 PR tree-optimization/82436
1268 * tree-vect-slp.c (vect_supported_load_permutation_p): More
1269 conservatively choose the vectorization factor when checking
1270 whether we can perform the required load permutation.
1271 (vect_transform_slp_perm_load): Assert when we may not fail.
1273 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
1275 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
1276 message for incompatible -msdata=* and -mcall-* options.
1278 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
1280 * i386.c (ia32_multipass_dfa_lookahead): Default to issue rate
1281 for post-reload scheduling.
1283 2017-10-05 Tamar Christina <tamar.christina@arm.com>
1285 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
1287 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
1289 * i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
1290 to improve monte carlo in scimark.
1292 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
1294 * i386.c (ix86_size_cost, i386_cost, i486_cost, pentium_cost,
1295 lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
1296 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
1297 pentium4_cost, nocona_cost): Set reassociation width to 1.
1298 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
1299 width to 2 for fp operations and 1 otherwise.
1300 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
1302 (atom_cost): Set reassociation width to 2.
1303 (slm_cost, generic_cost): Set fp reassociation width
1304 to 2 and 1 otherwise.
1305 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
1306 (core_cost): Set fp reassociation width to 4 and vector to 2.
1307 (ix86_reassociation_width): Rewrite using cost table; special case
1308 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
1309 and TARGET_AVX128_OPTIMAL.
1310 * i386.h (processor_costs): Add
1311 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
1312 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
1313 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
1314 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
1315 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
1316 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
1318 2017-10-05 Nathan Sidwell <nathan@acm.org>
1320 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
1322 2017-10-05 Tamar Christina <tamar.christina@arm.com>
1324 * config/arm/arm.c (arm_test_fpu_data): New.
1325 (arm_run_selftests): Call arm_test_fpu_data.
1327 2017-10-04 Nathan Sidwell <nathan@acm.org>
1329 * toplev.c (toplev::main): Remove excess parens on pretty_printer
1331 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
1333 2017-10-04 Sudakshina Das <sudi.das@arm.com>
1335 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
1336 check type for aarch64_simd_valid_immediate.
1337 (aarch64_output_simd_mov_immediate): Update prototype.
1338 (aarch64_simd_valid_immediate): Update prototype.
1339 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
1340 support for ORR-immediate.
1341 (and<mode>3): modified pattern to add support for BIC-immediate.
1342 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
1343 now checks for valid immediate for BIC and ORR based on new enum
1345 (aarch64_output_simd_mov_immediate): Function now used to output
1346 BIC/ORR imm as well based on new enum argument.
1347 * config/aarch64/constraints.md (Do): New vector immediate constraint.
1349 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
1350 (aarch64_reg_or_bic_imm): Likewise.
1352 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1354 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
1355 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
1357 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
1361 PR rtl-optimization/82396
1362 * haifa-sched.c (autopref_multipass_init): Simplify
1364 (autopref_rank_data): Simplify sort order.
1365 * sched-int.h (autopref_multipass_data_): Remove
1366 multi_mem_insn_p, min_offset and max_offset.
1368 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1370 * doc/sourcebuild.texi: Document vect_peeling_profitable.
1372 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1374 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
1377 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1379 * doc/sourcebuild.texi: Document vect_long_mult.
1381 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
1383 PR tree-optimization/82413
1384 * fold-const.c (build_range_check): Use widest_int when comparing
1385 the maximum ETYPE value with HIGH.
1387 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
1389 PR rtl-optimization/82396
1390 * haifa-sched.c (autopref_multipass_init): Simplify
1392 (autopref_rank_data): Simplify sort order.
1393 * sched-int.h (autopref_multipass_data_): Remove
1394 multi_mem_insn_p, min_offset and max_offset.
1396 2017-10-04 Jakub Jelinek <jakub@redhat.com>
1398 PR tree-optimization/82381
1399 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
1400 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
1403 PR tree-optimization/82374
1404 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
1405 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
1406 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
1407 current_function_decl to the new decl.
1409 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
1411 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
1412 helper macro for IEEE float128 hardware built-in functions.
1413 (SQRTF128_ODD): Add built-in functions with the round-to-odd
1415 (TRUNCF128_ODD): Likewise.
1416 (ADDF128_ODD): Likewise.
1417 (SUBF128_ODD): Likewise.
1418 (MULF128_ODD): Likewise.
1419 (DIVF128_ODD): Likewise.
1420 (FMAF128_ODD): Likewise.
1421 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
1422 UNSPEC_TRUNC_ROUND_TO_ODD.
1423 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
1424 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
1425 floating point round to odd instructions.
1426 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
1427 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
1428 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
1429 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
1430 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
1431 (trunc<mode>sf2_hw): Change the truncate with round to odd
1432 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
1433 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
1434 to odd hardware instructions.
1435 (sub<mode>3_odd): Likewise.
1436 (mul<mode>3_odd): Likewise.
1437 (div<mode>3_odd): Likewise.
1438 (sqrt<mode>2_odd): Likewise.
1439 (fma<mode>4_odd): Likewise.
1440 (fms<mode>4_odd): Likewise.
1441 (nfma<mode>4_odd): Likewise.
1442 (nfms<mode>4_odd): Likewise.
1443 (trunc<mode>df2_odd): Change the truncate with round to odd
1444 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
1446 * doc/extend.texi (PowerPC built-in functions): Update documentation
1447 for existing IEEE float128-bit built-in functions. Add built-in
1448 functions that generate the IEEE 128-bit floating point round to
1451 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
1453 PR rtl-optimization/77729
1454 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
1455 to (X&(C1&~C2))|C2 transformations.
1457 2017-10-03 Martin Jambor <mjambor@suse.cz>
1459 PR tree-optimization/82363
1460 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
1461 mismatch, mark lacc written regardless of racc.
1463 2017-10-03 Jakub Jelinek <jakub@redhat.com>
1465 PR tree-optimization/82381
1466 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
1467 stmt_to_insert nor wheather SSA_NAMEs are default defs.
1468 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
1469 fallthrough into reassoc_stmt_dominates_stmt_p.
1472 * combine.c (combine_instructions): Don't combine in unreachable
1475 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
1478 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
1479 function to not use the have_cpu variable. Do not set cpu_index,
1480 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
1482 (rs6000_valid_attribute_p): Remove duplicate initializations of
1483 old_optimize and func_optimize.
1484 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
1485 (rs6000_activate_target_options): Make global.
1486 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
1489 2017-10-02 Jakub Jelinek <jakub@redhat.com>
1491 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
1492 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
1493 Set *poffset to 0 and *psize and *pmax_size to -1 if
1494 *poffset + *psize overflows in HOST_WIDE_INT.
1496 PR tree-optimization/82387
1497 PR tree-optimization/82388
1498 PR tree-optimization/82389
1499 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
1500 instead of live_bytes non-NULL.
1502 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
1505 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
1508 2017-10-02 Richard Biener <rguenther@suse.de>
1510 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
1511 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
1512 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
1514 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
1516 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
1517 requested precision matches the type's.
1518 * calls.c (alloc_max_size): Calculate the new candidate size as
1519 a widest_int and use wi::to_widest when comparing it with the
1520 current candidate size.
1521 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
1522 zero rather than integer_zero_node.
1523 * match.pd: Check for a no-op conversion before using wi::add
1524 rather than after. Use tree_to_uhwi when summing small shift
1525 counts into an unsigned int.
1527 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
1528 Alan Hayward <alan.hayward@arm.com>
1529 David Sherwood <david.sherwood@arm.com>
1532 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
1533 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
1534 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
1535 POINTER_AND_FP_REGS.
1537 2017-10-02 Richard Biener <rguenther@suse.de>
1539 PR tree-optimization/82355
1540 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
1541 a mapping for the enclosing loop but avoid generating one for
1543 (copy_bb_and_scalar_dependences): Remove premature codegen
1544 error on PHIs in blocks duplicated into multiple places.
1545 * graphite-scop-detection.c
1546 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
1547 in the region use it as loop and nest to analyze the DR in.
1548 (try_generate_gimple_bb): Likewise.
1549 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
1550 (add_loop_constraints): For blocks in a loop not in the region
1551 create a dimension with a single iteration.
1552 * sese.h (gbb_loop_at_index): Remove assert.
1554 2017-10-01 Kevin Buettner <kevinb@redhat.com>
1556 * omp-expand.c (adjust_context_scope): New function.
1557 (expand_parallel_call): Call adjust_context_scope.
1559 2017-10-01 Jeff Law <law@redhat.com>
1561 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
1562 dom_opt_dom_walker class with direct access to private members.
1563 Add comments. Call test_for_singularity.
1564 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
1565 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
1566 m_dummy_cond anymore.
1567 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
1569 (pass_dominator:execute): Build the dummy_cond here and pass it
1570 to the dom_opt_dom_walker ctor.
1571 (test_for_singularity): New function.
1573 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
1574 Maya Rashish <coypu@sdf.org>
1576 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
1577 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
1578 (alpha*-*-netbsd*) Use nbsd_tm_file.
1579 (arm*-*-netbsdelf*) Likewise.
1580 (i[34567]86-*-netbsdelf*) Likewise.
1581 (x86_64-*-netbsd*) Likewise.
1582 (mips*-*-netbsd*) Likewise.
1583 (powerpc-*-netbsd*) Likewise.
1584 (sh*-*-netbsd*) Likewise.
1585 (sparc-*-netbsdelf*) Likewise.
1586 (sparc64-*-netbsd*) Likewise.
1587 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
1589 (vax-*-netbsdelf*) Likewise.
1590 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
1591 (UINT_FAST8_TYPE) Likewise.
1592 (INT_FAST16_TYPE) Check CHAR_FAST16.
1593 (UINT_FAST16_TYPE) Likewise.
1595 2017-09-30 Jakub Jelinek <jakub@redhat.com>
1598 * config/i386/i386.md
1599 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
1600 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
1601 *divmodsi4_zext_2): New define_insn_and_split.
1602 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
1603 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
1604 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
1605 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
1606 New define_insn_and_split.
1607 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
1608 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
1609 operands[1] having DImode when mode is SImode.
1611 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
1612 always SImode for DIV and MOD in REG_EQUAL notes.
1614 2017-09-29 Yury Gribov <tetra2005@gmail.com>
1617 * match.pd: Fix handling of NaNs in pattern.
1619 2017-09-29 Jeff Law <law@redhat.com>
1621 * sbitmap.c (bitmap_bit_in_range_p): New function.
1622 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
1623 * tree-ssa-dse.c (live_bytes_read): New function.
1624 (dse_classify_store): Ignore reads of dead bytes.
1626 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
1627 typos and whitespace errors.
1628 * config/i386/predicates.md (address_no_seg_operand): Likewise.
1629 * config/s390/s390.c (s390_emit_prologue): Likewise.
1631 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
1634 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
1635 with a symbol for LRA.
1637 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
1639 PR rtl-optimization/82338
1640 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
1642 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
1644 * genmodes.c (calc_wider_mode): Suppress qsort macro.
1645 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
1646 (qsort_chk): Declare.
1647 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
1648 (qsort_chk): New function.
1650 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1652 PR tree-optimization/82337
1653 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
1654 phi definition if the PHI result appears in an abnormal PHI.
1655 (find_basis_for_base_expr): Don't record a basis if the LHS of the
1656 basis appears in an abnormal PHI.
1658 2017-09-29 Richard Biener <rguenther@suse.de>
1660 * graphite-isl-ast-to-gimple.c
1661 (translate_isl_ast_to_gimple::set_codegen_error): New function.
1662 (binary_op_to_tree): Use it.
1663 (get_rename_from_scev): Likewise.
1664 (copy_loop_phi_nodes): Likewise.
1665 (copy_bb_and_scalar_dependences): Likewise.
1666 (translate_pending_phi_nodes): Likewise.
1668 2017-09-29 Jakub Jelinek <jakub@redhat.com>
1671 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
1672 for movabsq $(i32 << shift), r64.
1674 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
1676 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
1677 index when encoding %esp as %rsp to avoid 0x67 prefix.
1679 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
1681 * config/i386/i386.md (*movsf_internal, *movdf_internal):
1682 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
1684 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
1686 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
1687 Extensions with more than 16 double VFP registers.
1688 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
1689 to_clear_mask and all code related to it. Replace the remaining
1690 entry by a sbitmap and adapt code accordingly.
1692 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
1694 * brig-builtins.def: Change pure attributes to const.
1696 2017-09-28 Joseph Myers <joseph@codesourcery.com>
1698 * config.gcc (default_gnu_indirect_function): Default to yes for
1699 sparc*-*-linux* with glibc.
1701 2017-09-28 Joseph Myers <joseph@codesourcery.com>
1703 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
1704 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
1705 when creating .init_array and .fini_array sections with priority
1708 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
1711 * config/aarch64/aarch64.c
1712 (aarch64_builtin_support_vector_misalignment): Always return false
1713 when misalignment is unknown.
1715 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
1717 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
1718 this function to return false if the definition used by the swap
1719 instruction is artificial, or if the memory address from which the
1720 constant value is loaded is not represented by a base address held
1721 in a register or if the base address register is a frame or stack
1722 pointer. Additionally, return false if the base address of the
1723 loaded constant is a SYMBOL_REF but is not considered to be a
1725 (replace_swapped_load_constant): New function.
1726 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
1727 loaded constant vector with a load of a swapped constant vector.
1729 2017-09-27 Carl Love <cel@us.ibm.com>
1731 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
1732 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
1733 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
1736 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
1738 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
1739 first, always call autopref_rank_data otherwise.
1741 2017-09-27 Richard Biener <rguenther@suse.de>
1743 * graphite-scop-detection.c (find_scop_parameters): Move
1744 loop bound handling ...
1745 (gather_bbs::before_dom_children): ... here, avoiding the need
1746 to build scop_info->loop_nest.
1747 (record_loop_in_sese): Remove.
1748 * sese.h (sese_info_t::loop_nest): Remove.
1749 * sese.c (new_sese_info): Do not allocate loop_nest.
1750 (free_sese_info): Do not free loop_nest.
1752 2017-09-27 Jakub Jelinek <jakub@redhat.com>
1755 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
1756 lhs from calls if the lhs has addressable type.
1758 2017-09-27 Richard Biener <rguenther@suse.de>
1760 * graphite.h (scop::max_alias_set): New member.
1761 * graphite-scop-detection.c: Remove references to non-existing
1762 --param in comments.
1763 (build_alias_sets): Record the maximum alias set used for drs.
1764 (build_scops): Support zero as unlimited for
1765 --param graphite-max-arrays-per-scop.
1766 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
1768 (build_poly_sr_1): ... here. Compute alias set based on the
1769 maximum alias set used for drs rather than
1770 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
1772 2017-09-27 Richard Biener <rguenther@suse.de>
1774 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
1775 --param loop-block-tile-size=0 to disable tiling.
1777 2017-09-27 Richard Biener <rguenther@suse.de>
1779 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
1780 (graphite-max-nb-scop-params): Document special value zero.
1781 * domwalk.h (dom_walker::STOP): New symbolical constant.
1782 (dom_walker::dom_walker): Add optional parameter for bb to
1784 (dom_walker::~dom_walker): Declare.
1785 (dom_walker::before_dom_children): Document STOP return value.
1786 (dom_walker::m_user_bb_to_rpo): New member.
1787 (dom_walker::m_bb_to_rpo): Likewise.
1788 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
1789 mapping here if not provided by the user.
1790 (dom_walker::~dom_walker): Free bb to RPO mapping if not
1791 provided by the user.
1792 (dom_walker::STOP): Define.
1793 (dom_walker::walk): Do not compute bb to RPO mapping here.
1794 Support STOP return value from before_dom_children to stop
1796 * graphite-optimize-isl.c (optimize_isl): If the schedule
1797 is the same still generate code if -fgraphite-identity
1798 or -floop-parallelize-all are given.
1799 * graphite-scop-detection.c: Include cfganal.h.
1800 (gather_bbs::gather_bbs): Get and pass through bb to RPO
1802 (gather_bbs::before_dom_children): Return STOP for BBs
1804 (build_scops): Compute bb to RPO mapping and pass it to
1805 the domwalk. Treat --param graphite-max-nb-scop-params=0
1806 as not limiting the number of params.
1807 * graphite.c (graphite_initialize): Remove limit on the
1808 number of basic-blocks in a function.
1809 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
1810 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
1811 default value of 10.
1813 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
1815 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
1816 Adjust code to eliminate needing to do the shift right 32-bits
1817 operation after XSCVDPSPN.
1819 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1821 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
1822 ((X / Y) != 0 -> X >= Y): Likewise.
1824 2017-09-26 Carl Love <cel@us.ibm.com>
1826 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
1827 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
1828 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
1829 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
1830 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
1831 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
1832 definitions and overloading.
1833 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
1834 statement for P9V_BUILTIN_XST_LEN_R.
1835 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
1836 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
1837 define_expand and define_insn for the instructions and builtins.
1838 * doc/extend.texi: Update the built-in documentation file for the new
1840 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
1841 define_insn for the instructions
1843 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
1846 * gcc/config/netbsd-protos.h: New file.
1847 * gcc/config/netbsd.c: New file.
1848 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
1849 * gcc/config/t-netbsd: New file.
1850 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
1851 (tmake_file) Add t-netbsd.
1852 (extra_objs) Add netbsd.o.
1854 2017-09-26 Janus Weil <janus@gcc.gnu.org>
1858 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
1860 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
1862 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
1863 sign extension from a vector register to a GPR by doing a 32-bit
1864 direct move and then an EXTSW.
1865 (extendsi<mode>2 splitter): Likewise.
1866 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
1867 right or vector extract after doing XSCVDPSPN. Use
1868 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
1870 (movdi_from_sf_zero_ext): Likewise.
1871 (reload_gpr_from_vsxsf): Likewise.
1872 (p8_mfvsrd_4_disf): Delete, no longer used.
1873 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
1874 then needing to move the SFmode to a GPR to use the XSCVDPSP
1875 instruction instead of FRSP and XSCVDPSPN.
1876 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
1877 it is adjacent to the other XSCVSPDP insns.
1878 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
1879 SFmode to be in traditional Altivec registers.
1880 (vsx_xscvdpspn): Eliminate useless alternative constraint.
1881 (vsx_xscvspdpn): Likewise.
1882 (vsx_xscvspdpn_scalar): Likewise.
1884 2017-09-26 Martin Jambor <mjambor@suse.cz>
1886 * tree-sra.c (compare_access_positions): Put integral types first,
1887 stabilize sorting of integral types, remove conditions putting
1888 non-full-precision integers last.
1889 (sort_and_splice_var_accesses): Disable scalarization if a
1890 non-integert would be represented by a non-full-precision integer.
1892 2017-09-26 Joseph Myers <joseph@codesourcery.com>
1894 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
1895 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
1896 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
1897 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
1898 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
1899 conditionals inside the function instead of around it. Call
1900 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
1901 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
1903 2017-09-26 Richard Biener <rguenther@suse.de>
1905 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
1907 (scop_detection::build_scop_breadth): ... this. Removed.
1908 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
1909 (scop_detection::harmful_stmt_in_bb): Likewise.
1910 (scop_detection::graphite_can_represent_stmt): Likewise.
1911 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
1912 (scop_detection::can_represent_loop): Remove recursion, fold in ...
1913 (scop_detection::can_represent_loop_1): ... this. Removed.
1914 (scop_detection::harmful_loop_in_region): Simplify after inlining
1915 the above and remove more quadraticness.
1916 (build_scops): Adjust.
1917 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
1920 2017-09-26 Jakub Jelinek <jakub@redhat.com>
1923 * config/i386/i386.c (ix86_print_operand_address_as): Only test
1924 REGNO (base) == SP_REG if base is a REG.
1927 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
1928 if it is different SSA_NAME.
1929 (optimize_range_tests_cmp_bitwise): New function.
1930 (optimize_range_tests): Call it.
1932 2017-09-26 Richard Biener <rguenther@suse.de>
1934 PR tree-optimization/82321
1935 * graphite.c (canonicalize_loop_closed_ssa): Properly check
1936 for the def being inside the loop.
1938 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1940 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
1942 * config/s390/s390-builtins.def: Fix constraint on op4.
1944 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1946 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
1947 independent expanders.
1948 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
1949 ("vec_ordered", "vec_unordered"): New expanders.
1951 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1953 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
1956 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1958 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
1959 vec_unpacks_lo_v16qi.
1960 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
1962 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1964 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
1965 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
1966 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
1968 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1970 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
1972 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
1974 ("*vec_slb<mode>"): New insn pattern.
1975 ("vec_shr_<mode>"): New expander.
1976 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
1977 and force the shift count operand to V16QImode.
1978 ("vec_srb<mode>"): Set shift count mode to V16QI.
1980 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1982 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
1983 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
1984 ("vec_widen_smult_hi_<mode>"): New expander definitions.
1986 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
1989 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
1991 2017-09-26 Richard Biener <rguenther@suse.de>
1993 PR tree-optimization/82320
1994 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
1997 2017-09-25 Jeff Law <law@redhat.com>
1999 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
2000 prototype for new argument.
2001 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
2002 mostly extracted from rs6000_emit_allocate_stack.
2003 (rs6000_emit_probe_stack_range_stack_clash): New function.
2004 (rs6000_emit_allocate_stack): Call
2005 rs6000_emit_probe_stack_range_stack_clash as needed.
2006 (rs6000_emit_probe_stack_range): Add additional argument
2007 to call to gen_probe_stack_range{si,di}.
2008 (output_probe_stack_range): New.
2009 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
2010 (output_probe_stack_range_stack_clash): New.
2011 (rs6000_emit_prologue): Emit notes into dump file as requested.
2012 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
2013 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
2014 Add additional operand and pass it to output_probe_stack_range.
2016 2017-09-25 Bin Cheng <bin.cheng@arm.com>
2018 PR tree-optimization/82163
2019 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
2020 (checking_verify_loop_closed_ssa): New parameter.
2021 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
2022 (check_loop_closed_ssa_stmt): Delete.
2023 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
2024 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
2025 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
2028 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
2030 * brig-builtins.def: Treat HSAIL barrier builtins as
2031 setjmp/longjump style functions.
2033 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
2035 * target.def (constant_alignment): New hook.
2036 * defaults.h (CONSTANT_ALIGNMENT): Delete.
2037 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
2038 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
2039 * doc/tm.texi: Regenerate.
2040 * targhooks.h (default_constant_alignment): Declare.
2041 (constant_alignment_word_strings): Likewise.
2042 * targhooks.c (default_constant_alignment): New function.
2043 (constant_alignment_word_strings): Likewise.
2044 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
2045 instead of CONSTANT_ALIGNMENT.
2046 * varasm.c (align_variable, get_variable_align, build_constant_desc)
2047 (force_const_mem): Likewise.
2048 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
2049 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
2050 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
2051 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2052 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
2054 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
2055 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2056 constant_alignment_word_strings.
2057 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
2058 (CONSTANT_ALIGNMENT): Likewise.
2059 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2060 (arm_constant_alignment): New function.
2061 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
2062 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2063 constant_alignment_word_strings.
2064 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
2065 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2066 constant_alignment_word_strings.
2067 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
2068 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2069 (cris_constant_alignment): New function.
2070 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
2071 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2072 (epiphany_constant_alignment): New function.
2073 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
2074 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2075 constant_alignment_word_strings.
2076 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
2077 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2078 constant_alignment_word_strings.
2079 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
2080 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2081 constant_alignment_word_strings.
2082 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
2083 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
2084 * config/i386/i386.c (ix86_constant_alignment): Make static.
2085 Use the same interface as the target hook.
2086 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2087 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
2088 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2089 constant_alignment_word_strings.
2090 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
2091 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
2092 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2093 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
2094 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2095 constant_alignment_word_strings.
2096 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
2097 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2098 constant_alignment_word_strings.
2099 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
2100 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2101 constant_alignment_word_strings.
2102 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
2103 * config/microblaze/microblaze.c (microblaze_constant_alignment):
2105 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2106 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
2107 * config/mips/mips.c (mips_constant_alignment): New function.
2108 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2109 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
2110 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
2111 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2112 (mmix_constant_alignment): Make static. Use the same interface
2114 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
2115 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2116 constant_alignment_word_strings.
2117 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
2118 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2119 constant_alignment_word_strings.
2120 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
2121 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2122 constant_alignment_word_strings.
2123 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
2124 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2125 (rs6000_constant_alignment): New function.
2126 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
2127 * config/riscv/riscv.c (riscv_constant_alignment): New function.
2128 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2129 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
2130 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2131 (rs6000_constant_alignment): New function.
2132 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
2133 * config/s390/s390.c (s390_constant_alignment): New function.
2134 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2135 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
2136 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2137 constant_alignment_word_strings.
2138 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
2139 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2140 (sparc_constant_alignment): New function.
2141 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
2142 * config/spu/spu.c (spu_constant_alignment): New function.
2143 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2144 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
2145 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2146 constant_alignment_word_strings.
2147 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
2148 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2149 constant_alignment_word_strings.
2150 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
2151 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2152 constant_alignment_word_strings.
2153 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
2154 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2155 (visium_constant_alignment): New function.
2156 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
2157 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2158 (xtensa_constant_alignment): New function.
2159 * system.h (CONSTANT_ALIGNMENT): Poison.
2161 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
2163 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
2164 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
2165 (rs6000_builtin_valid_without_lhs): New helper function.
2166 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2167 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
2169 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
2171 * target.h (vec_perm_indices): Use unsigned short rather than
2173 (auto_vec_perm_indices): Likewise.
2174 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
2175 Use unsigned int rather than unsigned char.
2176 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
2178 2017-09-25 Richard Biener <rguenther@suse.de>
2180 * cfgloop.h (sort_sibling_loops): Declare.
2181 * cfgloop.c (sort_sibling_loops_cmp): New helper.
2182 (sort_sibling_loops): New function sorting the sibling loop list
2184 * graphite.c (graphite_transform_loops): Sort sibling loops.
2186 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
2188 * target.def (vec_perm_const_ok): Change sel parameter to
2190 * optabs-query.c (can_vec_perm_p): Update accordingly.
2191 * doc/tm.texi: Regenerate.
2192 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
2193 auto_vec_perm_indices and remove separate nelt field.
2194 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
2195 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
2196 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
2197 (aarch64_expand_vec_perm_const): Update accordingly.
2198 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
2199 to vec_perm_indices.
2200 * config/arm/arm.c (expand_vec_perm_d): Change perm to
2201 auto_vec_perm_indices and remove separate nelt field.
2202 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
2203 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
2204 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
2206 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
2207 to vec_perm_indices.
2208 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
2209 sel to vec_perm_indices.
2210 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
2211 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
2212 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
2214 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
2217 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
2220 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
2221 on the FUNCTION_DECL function context if it has a DIE that is a
2224 2017-09-25 Richard Biener <rguenther@suse.de>
2226 PR tree-optimization/82285
2227 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
2230 2017-09-25 Tom de Vries <tom@codesourcery.com>
2234 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
2237 2017-09-25 Richard Biener <rguenther@suse.de>
2239 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
2240 ISL errors other than isl_error_quota happen. Dump if the
2241 schedule is the same.
2242 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
2243 errors instead of aborting inside ISL.
2245 2017-09-25 Iain Sandoe <iain@codesourcery.com>
2248 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
2249 of libgcc_eh for m64.
2250 * config/i386/darwin64.h: Likewise.
2252 2017-09-25 Richard Biener <rguenther@suse.de>
2255 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
2256 attribute for incomplete types nor twice for complete ones.
2258 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
2261 * config/i386/i386.c (ix86_print_operand_address_as): Encode
2262 %esp as %rsp to avoid 0x67 prefix if there is no index or base
2265 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
2268 * config/i386/i386.opt (mprefer-avx256): Use
2269 ix86_target_flags variable.
2270 * config/i386/i386.c (ix86_target_string): Move
2271 -mprefer-avx256 to flag2_opts.
2273 2017-09-22 Jakub Jelinek <jakub@redhat.com>
2276 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
2277 and x != -1 | y != -1 into (x & y) != -1.
2279 2017-09-22 Steve Ellcey <sellcey@cavium.com>
2281 * config.gcc: Add new case statement to set
2282 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
2283 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
2284 s390x-*-linux* case statements. Added aarch64 to the list of
2285 supported architectures.
2287 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2289 PR tree-optimization/82289
2290 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
2291 STMT_VINFO_RELEVANT_P.
2293 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2294 Alan Hayward <alan.hayward@arm.com>
2295 David Sherwood <david.sherwood@arm.com>
2297 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
2298 for VR_RANGE only; don't allow VR_ANTI_RANGE.
2299 (extract_range_from_binary_expr_1): Don't call
2300 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
2302 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2303 Alan Hayward <alan.hayward@arm.com>
2304 David Sherwood <david.sherwood@arm.com>
2306 * target.def (preferred_vector_alignment): New hook.
2307 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
2309 * doc/tm.texi: Regenerate.
2310 * targhooks.h (default_preferred_vector_alignment): Declare.
2311 * targhooks.c (default_preferred_vector_alignment): New function.
2312 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
2314 (DR_TARGET_ALIGNMENT): New macro.
2315 (aligned_access_p): Update commentary.
2316 (vect_known_alignment_in_bytes): New function.
2317 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
2319 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
2320 Calculate the misalignment based on the target alignment rather than
2322 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
2323 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
2324 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
2325 the target alignment, rather than masking the element misalignment
2326 with the number of elements in a vector. Also use the target
2327 alignment when calculating the maximum number of peels.
2328 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
2329 instead of TYPE_ALIGN_UNIT.
2330 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
2331 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
2332 (vect_create_addr_base_for_vector_ref): Update call accordingly.
2333 (vect_create_data_ref_ptr): Likewise.
2334 (vect_setup_realignment): Realign by ANDing with
2335 -DR_TARGET_MISALIGNMENT.
2336 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
2337 the number of peels based on DR_TARGET_ALIGNMENT.
2338 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
2339 with the guaranteed alignment boundary when deciding whether
2341 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
2342 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
2343 (ensure_base_align): Remove stmt_info parameter. Get the
2344 target base alignment from DR_TARGET_ALIGNMENT.
2345 (vectorizable_store): Update call accordingly. Interpret
2346 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
2348 (vectorizable_load): Likewise.
2350 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2351 Alan Hayward <alan.hayward@arm.com>
2352 David Sherwood <david.sherwood@arm.com>
2354 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
2355 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
2356 (vect_enhance_data_refs_alignment): Likewise.
2358 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
2360 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
2361 error. Only quit immediately if parsing is complete.
2362 (BEGIN): Initialize fatal_err and parse_done.
2363 (begin fpu, end fpu): Check number of arguments.
2364 (begin arch, end arch): Likewise.
2365 (begin cpu, end cpu): Likewise.
2366 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
2367 (optalias): Likewise.
2369 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
2371 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
2372 * config/arm/arm-isa.h: Delete. Move definitions to ...
2373 * arm-cpus.in: ... here. Use new feature and fgroup values.
2374 * config/arm/arm.c (arm_option_override): Use lower case for feature
2376 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
2377 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
2378 * config/arm/parsecpu.awk (END): Add new command 'isa'.
2380 (print_isa_bits_for): New function.
2381 (gen_isa): New function.
2382 (gen_comm_data): Use print_isa_bits_for.
2383 (define feature): New keyword.
2384 (define fgroup): New keyword.
2385 * config/arm/t-arm (TM_H): Remove.
2386 (GTM_H): Add arm-isa.h.
2387 (arm-isa.h): Add rule to generate file.
2388 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
2389 case for feature bit names.
2391 2017-09-22 Richard Biener <rguenther@suse.de>
2393 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
2395 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
2396 print of no dependency loops ...
2397 * graphite.c (graphite_transform_loops): ... here.
2398 (canonicalize_loop_closed_ssa_form): Work from inner to outer
2400 (same_close_phi_node, remove_duplicate_close_phi,
2401 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
2402 (canonicalize_loop_closed_ssa): ... here and simplify.
2403 * graphite-optimize-isl.c: Include tree-vectorizer.h.
2404 (optimize_isl): Use dump_printf_loc to tell when we stopped
2405 optimizing because of an ISL timeout.
2407 2017-09-22 Richard Biener <rguenther@suse.de>
2409 PR tree-optimization/82291
2410 * tree-if-conv.c (predicate_mem_writes): Make sure to
2411 remove writes in blocks predicated with false.
2413 2017-09-22 Richard Biener <rguenther@suse.de>
2415 * sese.c: Include cfganal.h.
2416 (if_region_set_false_region): Remove.
2417 (create_if_region_on_edge): Likewise.
2418 (move_sese_in_condition): Re-implement without destroying
2421 2017-09-22 Richard Biener <rguenther@suse.de>
2423 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
2424 Verify both BBs contain loop PHI nodes before dispatching to
2426 (graphite_regenerate_ast_isl): Do not recompute dominators,
2427 do not verify three times. Restructure for clarity.
2428 * graphite-scop-detection.c (same_close_phi_node,
2429 remove_duplicate_close_phi, make_close_phi_nodes_unique,
2430 defined_in_loop_p, canonicalize_loop_closed_ssa,
2431 canonicalize_loop_closed_ssa_form): Simplify, remove excess
2432 checking and SSA rewrite, move to ...
2433 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
2434 (graphite_initialize): Do not pass in ctx, do not reset the
2435 SCEV cache, compute only dominators.
2436 (graphite_transform_loops): Allocate ISL ctx after
2437 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
2438 Maintain post-dominators only around build_scops.
2439 * sese.c (if_region_set_false_region): Make static. Free
2440 and recompute dominators.
2441 (move_sese_in_condition): Assert we don't get called with
2442 post-dominators computed.
2443 * sese.h (if_region_set_false_region): Remove.
2445 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
2447 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
2448 mode attribute for TARGET_AVX512VL.
2450 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
2452 * config/i386/i386.opt (mprefer-avx256): New option.
2453 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
2455 (ix86_preferred_simd_mode): Return 256-bit AVX modes
2456 for TARGET_PREFER_AVX256.
2457 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
2459 2017-09-21 Jeff Law <law@redhat.com>
2461 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
2462 Fix dump output if the only stack space is for pushed registers.
2464 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2466 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
2469 2017-09-21 Martin Sebor <msebor@redhat.com>
2472 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
2473 code (in C++) or code that triggers warnings.
2475 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
2477 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
2479 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
2481 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
2482 * sched-rgn.c: Ditto.
2483 * sel-sched-ir.c: Ditto.
2485 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
2487 * toplev.h (set_random_seed): Adjust return type.
2488 * toplev.c (init_local_tick): Move eager initialization of random_seed
2489 to get_random_seed. Adjust comment.
2490 (init_random_seed): Inline to get_random_seed, delete.
2491 (get_random_seed): Initialize random_seed lazily.
2492 (set_random_seed): Do not return previous value.
2493 (print_switch_value): Do not call get_random_seed.
2495 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
2497 * cgraph.c (delete_function_version): New, broken out from...
2498 (cgraph_node::delete_function_version): ...here. Rename to
2499 cgraph_node::delete_function_version_by_decl. Update all uses.
2500 (cgraph_node::remove): Call delete_function_version.
2502 2017-09-21 Jakub Jelinek <jakub@redhat.com>
2505 * tree-inline.c (expand_call_inline): Emit clobber stmts for
2506 VAR_DECLs to which addressable non-volatile parameters are mapped
2507 and for id->retvar after the return value assignment. Clear
2508 id->retval and id->retbnd after inlining.
2510 2017-09-21 Richard Biener <rguenther@suse.de>
2512 PR tree-optimization/82276
2513 PR tree-optimization/82244
2514 * tree-vrp.c (build_assert_expr_for): Set
2515 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
2517 (remove_range_assertions): Revert earlier change.
2519 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
2522 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
2524 2017-09-21 Richard Biener <rguenther@suse.de>
2526 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
2527 Restore valid IL after code generation errors.
2528 * graphite.c (graphite_transform_loops): Diagnose code
2529 generation issues as MSG_MISSED_OPTIMIZATION and continue
2530 with processing SCOPs.
2532 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2533 Alan Hayward <alan.hayward@arm.com>
2534 David Sherwood <david.sherwood@arm.com>
2536 * calls.c (compute_argument_addresses): Use simplify_gen_binary
2537 rather than choosing between plus_constant and gen_rtx_<CODE>.
2538 * expr.c (emit_push_insn): Likewise.
2539 (expand_expr_real_2): Likewise.
2541 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2542 Alan Hayward <alan.hayward@arm.com>
2543 David Sherwood <david.sherwood@arm.com>
2545 * loop-unroll.c (split_iv): Call copy_rtx on the step.
2547 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2548 Alan Hayward <alan.hayward@arm.com>
2549 David Sherwood <david.sherwood@arm.com>
2551 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
2552 calling tree_to_uhwi.
2554 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2555 Alan Hayward <alan.hayward@arm.com>
2556 David Sherwood <david.sherwood@arm.com>
2558 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
2559 INTEGER_CST rather than a negative test for ADDR_EXPR.
2561 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2562 Alan Hayward <alan.hayward@arm.com>
2563 David Sherwood <david.sherwood@arm.com>
2565 * tree-vrp.c (extract_range_from_binary_expr_1): Check
2566 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
2568 2017-09-21 Richard Biener <rguenther@suse.de>
2570 PR tree-optimization/71351
2571 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
2572 graphite_create_new_loop_guard): Remove, fold remaining parts
2574 (translate_isl_ast_node_for): ... here and simplify.
2576 2017-09-21 Jakub Jelinek <jakub@redhat.com>
2579 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
2580 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
2581 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
2582 alternative always use QI mode, for -Os imov (=R,R) alternative
2583 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
2586 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2587 Jeff Law <law@redhat.com>
2589 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
2590 (allocate_stack_space): New function, partially extracted from
2592 (s390_emit_prologue): Track offset to most recent stack probe.
2593 Code to allocate space moved into allocate_stack_space.
2594 Dump actions when no stack is allocated.
2595 (s390_prologue_plus_offset): New function.
2596 (s390_emit_stack_probe): Likewise.
2598 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
2600 * common.opt (Wa, Wl, Wp, g, gz=): Add
2602 (gno-column-info): Remove.
2603 (gcolumn-info): Drop RejectNegative.
2605 (gno-record-gcc-switches): Remove.
2606 (grecord-gcc-switches): Drop RejectNegative.
2607 (gno-split-dwarf): Remove.
2608 (gsplit-dwarf): Drop RejectNegative.
2609 (gno-strict-dwarf): Remove.
2610 (gstrict-dwarf): Drop RejectNegative.
2611 * config/darwin.opt (gfull, gused): Add RejectNegative.
2612 * dwarf2out.c (gen_producer_string): Drop
2613 gno-record-gcc-switches handler.
2614 * optc-gen.awk: Add g to prefixes with negative forms.
2615 * opts-common.c (remapping_prefix_p): New.
2616 (find_opt): Check it.
2617 (generate_canonical_option): Test g prefix.
2618 (option_map): Add -gno- mapping.
2619 (add_misspelling_candidates): Check remapping_prefix_p.
2621 2017-09-20 Jeff Law <law@redhat.com>
2623 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
2624 thinko in stack clash protection support.
2626 * explow.c (compute_stack_clash_protection_loop_data): Use
2627 CONST_INT_P instead of explicit test. Verify object is a
2628 CONST_INT_P before looking at INTVAL.
2629 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
2630 instead of explicit test.
2632 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
2635 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
2636 address instead of to r1 and r11.
2638 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
2640 * config.gcc: Support "knm".
2641 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
2642 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2644 * config/i386/i386.c (m_KNM): Define.
2645 (processor_target_table): Add "knm".
2647 (ix86_option_override_internal): Add "knm".
2648 (ix86_issue_rate): Add PROCESSOR_KNM.
2649 (ix86_adjust_cost): Ditto.
2650 (ia32_multipass_dfa_lookahead): Ditto.
2651 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
2652 (fold_builtin_cpu): Add M_INTEL_KNM.
2653 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
2654 (processor_type): Add PROCESSOR_KNM.
2655 * config/i386/x86-tune.def: Add m_KNM.
2656 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
2658 2017-09-20 Richard Biener <rguenther@suse.de>
2660 PR tree-optimization/80213
2661 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
2662 are allowed in empty BBs as well.
2663 (canonicalize_loop_closed_ssa): Also look for other complex
2665 (scop_detection::get_sese): Include the loop-closed PHI block
2667 (scop_detection::merge_sese): Remove code adding extra blocks.
2668 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
2669 (build_scops): Assert the final returned scop is invalid.
2671 2017-09-20 Richard Biener <rguenther@suse.de>
2673 PR tree-optimization/82264
2674 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
2676 (vn_phi_lookup): Likewise.
2677 (vn_phi_insert): Likewise.
2679 2017-09-20 Jakub Jelinek <jakub@redhat.com>
2681 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
2682 that fits into uhwi or shwi, add DW_AT_const_value regardless
2683 of early_dwarf without going through RTL, using add_AT_unsigned
2686 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
2687 (DEBUG_LTO_ABBREV_SECTION): Likewise.
2688 (DEBUG_LTO_MACINFO_SECTION): Likewise.
2689 (DEBUG_MACRO_SECTION): Likewise.
2690 (DEBUG_LTO_MACRO_SECTION): Likewise.
2691 (DEBUG_STR_DWO_SECTION): Likewise.
2692 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
2693 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
2694 (DEBUG_LTO_DWO_LINE_SECTION): Define.
2695 (DEBUG_LTO_LINE_STR_SECTION): Define.
2696 (init_sections_and_labels): Initialize debug_line_str_section
2697 variable. Initialize debug_loc_section for -gdwarf-5 to
2698 DEBUG_LOCLISTS_SECTION. Formatting fixes.
2700 2017-09-20 Richard Biener <rguenther@suse.de>
2702 * graphite-sese-to-poly.c (extract_affine): Properly handle
2703 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
2705 2017-09-20 Richard Biener <rguenther@suse.de>
2707 PR tree-optimization/81373
2708 * graphite-scop-detection.c (build_cross_bb_scalars_def):
2709 Force SESE live-out defs to be handled even if they are
2712 2017-09-19 Jeff Law <law@redhat.com>
2714 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
2715 nothing for stack adjustments with REG_STACK_CHECK.
2716 * sched-deps.c (parse_add_or_inc): Reject insns with
2717 REG_STACK_CHECK from dependency breaking.
2718 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
2719 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
2720 * reg-notes.def (STACK_CHECK): New note.
2722 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
2723 (ix86_expand_prologue): Dump stack clash info as needed.
2724 Call ix86_adjust_stack_and_probe_stack_clash as needed.
2726 * function.c (dump_stack_clash_frame_info): New function.
2727 * function.h (dump_stack_clash_frame_info): Prototype.
2728 (enum stack_clash_probes): New enum.
2730 * config/alpha/alpha.c (alpha_expand_prologue): Also check
2731 flag_stack_clash_protection.
2732 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
2733 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
2734 (arm_frame_pointer_required): Likewise.
2735 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
2736 (ia64_expand_prologue): Likewise.
2737 * config/mips/mips.c (mips_expand_prologue): Likewise.
2738 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
2739 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
2740 (sparc_flat_expand_prologue): Likewise.
2741 * config/spu/spu.c (spu_expand_prologue): Likewise.
2743 * explow.c: Include "params.h".
2744 (anti_adjust_stack_and_probe_stack_clash): New function.
2745 (get_stack_check_protect): Likewise.
2746 (compute_stack_clash_protection_loop_data): Likewise.
2747 (emit_stack_clash_protection_loop_start): Likewise.
2748 (emit_stack_clash_protection_loop_end): Likewise.
2749 (allocate_dynamic_stack_space): Use get_stack_check_protect.
2750 Use anti_adjust_stack_and_probe_stack_clash.
2751 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
2752 (emit_stack_clash_protection_loop_start): Likewise.
2753 (emit_stack_clash_protection_loop_end): Likewise.
2754 * rtl.h (get_stack_check_protect): Prototype.
2755 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
2756 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
2757 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
2759 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
2761 * doc/tm.texi: Rebuilt.
2762 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
2763 get_stack_check_protect.
2764 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
2765 * config/arm/arm.c (arm_expand_prologue): Likewise.
2766 (arm_frame_pointer_required): Likewise.
2767 * config/i386/i386.c (ix86_expand_prologue): Likewise.
2768 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
2769 * config/mips/mips.c (mips_expand_prologue): Likewise.
2770 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
2771 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
2772 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
2773 (sparc_flat_expand_prologue): Likewise.
2775 * common.opt (-fstack-clash-protection): New option.
2776 * flag-types.h (enum stack_check_type): Note difference between
2777 -fstack-check= and -fstack-clash-protection.
2778 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
2779 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
2780 * toplev.c (process_options): Issue warnings/errors for cases
2781 not handled with -fstack-clash-protection.
2782 * doc/invoke.texi (-fstack-clash-protection): Document new option.
2783 (-fstack-check): Note additional problem with -fstack-check=generic.
2784 Note that -fstack-check is primarily for Ada and refer users
2785 to -fstack-clash-protection for stack-clash-protection.
2786 Document new params for stack clash protection.
2788 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
2790 * config/i386/i386.c (ix86_split_long_move): Do not handle
2791 address used for LEA in a special way.
2793 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
2795 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
2797 2017-09-19 Martin Sebor <msebor@redhat.com>
2800 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
2801 of incompatible types.
2803 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
2805 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
2806 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
2807 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2808 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
2810 2017-09-19 Richard Biener <rguenther@suse.de>
2812 PR tree-optimization/82244
2813 * tree-vrp.c (remove_range_assertions): Do not propagate
2814 a constant to abnormals but replace the assert with a copy.
2816 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
2818 PR rtl-optimization/57878
2819 PR rtl-optimization/68988
2820 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
2821 avoidance test involving non_reload_pseudos. Move frequency test
2822 below the general fragmentation avoidance test.
2824 2017-09-19 Richard Biener <rguenther@suse.de>
2826 PR tree-optimization/69728
2827 * graphite-sese-to-poly.c (schedule_error): New global.
2828 (add_loop_schedule): Handle empty domain by failing the
2830 (build_original_schedule): Handle schedule_error.
2832 2017-09-19 Richard Biener <rguenther@suse.de>
2834 * graphite-scop-detection.c (scop_detection::can_represent_loop):
2835 Do not iterate to sibling loops but only to siblings of inner
2838 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
2841 * config/m68k/m68k.md (moveq feeding equality comparison): Check
2842 that the registers are different.
2844 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
2846 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
2847 to processor_model and "amdfam17h" to arch_names_table.
2848 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
2850 2017-09-18 Jakub Jelinek <jakub@redhat.com>
2853 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
2855 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2856 Alan Hayward <alan.hayward@arm.com>
2857 David Sherwood <david.sherwood@arm.com>
2859 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
2861 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
2863 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
2864 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
2865 vect_schedule_slp_instance.
2866 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
2867 Update call to vect_slp_analyze_node_operations. Simplify return
2869 (vect_slp_analyze_bb_1): Update call accordingly.
2870 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
2871 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
2872 (vect_schedule_slp): Update call accordingly.
2874 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2875 Alan Hayward <alan.hayward@arm.com>
2876 David Sherwood <david.sherwood@arm.com>
2878 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
2879 with types that aren't in fact scalar.
2881 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2883 * tree-vect-slp.c (vect_record_max_nunits): New function,
2885 (vect_build_slp_tree_1): ...here.
2886 (vect_build_slp_tree_2): Call it for phis too.
2888 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2890 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
2891 to vect_get_vec_def_for_operand when getting the mask operand.
2893 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2894 Alan Hayward <alan.hayward@arm.com>
2895 David Sherwood <david.sherwood@arm.com>
2897 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
2900 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2901 Alan Hayward <alan.hayward@arm.com>
2902 David Sherwood <david.sherwood@arm.com>
2904 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
2905 calculation for vector booleans.
2907 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2908 Alan Hayward <alan.hayward@arm.com>
2909 David Sherwood <david.sherwood@arm.com>
2911 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
2913 (vect_transform_stmt): ...here.
2914 (vect_analyze_stmt): Use it instead of calling
2915 vectorizable_live_operation directly.
2917 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
2919 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
2920 non-SIMT targets in acc vector loops.
2922 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
2924 * configure.ac: Add arc and check if assembler supports gdwarf2.
2925 * configure: Regenerate.
2927 2017-09-18 Richard Biener <rguenther@suse.de>
2929 PR tree-optimization/82220
2930 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
2931 epilogue niters from the min_profitable_iters compute.
2933 2017-09-18 Jakub Jelinek <jakub@redhat.com>
2936 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
2937 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
2938 (ix86_init_pic_reg): Revert 2017-09-01 changes.
2940 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
2943 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
2944 switching to a new text section.
2946 2017-09-18 Richard Biener <rguenther@suse.de>
2948 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
2950 (build_alias_set): Reject aliases with no access function.
2952 2017-09-18 Richard Biener <rguenther@suse.de>
2954 PR tree-optimization/79622
2955 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
2957 (build_cross_bb_scalars_use): Likewise.
2959 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
2961 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
2963 2017-09-18 Alan Modra <amodra@gmail.com>
2966 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
2967 stack_pointer_rtx for count 0. Update comments. Break up
2968 large rtl expression.
2970 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
2972 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
2973 Increase to 20 bytes.
2974 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
2975 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
2976 or avx version of the stub.
2978 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
2981 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
2982 compute the minimum stack alignment. Also update preferred stack
2983 boundary for leaf functions.
2985 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
2987 PR tree-optimization/82228
2988 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
2991 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
2993 * common/config/nds32/nds32-common.c
2994 (nds32_option_optimization_table): Refine formatting.
2995 (nds32_option_optimization_table): Use -fsched-pressure and
2996 -fomit-frame-pointer for specific optimization level.
2998 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
3000 * config/nds32/nds32.c: Refine formatting and comments.
3001 * config/nds32/nds32.h: Likewise.
3002 * config/nds32/nds32.md: Likewise.
3003 * config/nds32/nds32-cost.c: Likewise.
3004 * config/nds32/nds32-isr.c: Likewise.
3005 * config/nds32/nds32-md-auxiliary.c: Likewise.
3006 * config/nds32/nds32-multiple.md: Likewise.
3007 * config/nds32/nds32-predicates.c: Likewise.
3009 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
3010 Jakub Jelinek <jakub@redhat.com>
3012 Add support for -std=c++2a.
3013 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
3015 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
3017 2017-09-15 Steve Ellcey <sellcey@cavium.com>
3020 * doc/extend.texi (Common Function Attributes): Add
3021 references to ARM, AArch64, and S/390 specific attributes.
3022 (Function Specific Option Pragmas): Add AArch64 and S/390
3023 to list of back ends that support the target pragma.
3025 2017-09-15 Nathan Sidwell <nathan@acm.org>
3027 * doc/standards.texi: Fix C++17 description. Update URLs for
3030 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
3032 * common.opt (Wcast-align=strict): New warning option.
3033 * doc/invoke.texi: Document -Wcast-align=strict.
3035 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
3037 * cgraph.h (cgraph_thunk_info): Add comments.
3038 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
3039 assert for VIRTUAL_* arguments stricter.
3041 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
3043 PR tree-optimization/71026
3044 * match.pd: Move RDIV patterns from fold-const.c
3045 * fold-const.c (distribute_real_division): Removed.
3046 (fold_binary_loc): Remove calls to distribute_real_divison.
3048 2017-09-15 Jakub Jelinek <jakub@redhat.com>
3050 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
3051 c++1z and gnu++1z as deprecated. Change other references to
3052 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
3053 Change -Wc++1z-compat to -Wc++17-compat.
3054 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
3055 * dwarf2out.c (highest_c_language): Handle C++17.
3056 (gen_compile_unit_die): Likewise.
3058 2017-09-15 Jakub Jelinek <jakub@redhat.com>
3060 PR rtl-optimization/82192
3061 * combine.c (make_extraction): Don't look through non-paradoxical
3062 SUBREGs or TRUNCATE if pos + len is or might be bigger than
3065 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
3066 Alan Hayward <alan.hayward@arm.com>
3067 David Sherwood <david.sherwood@arm.com>
3069 * target.def (function_arg_offset): New hook.
3070 * targhooks.h (default_function_arg_offset): Declare.
3071 * targhooks.c (default_function_arg_offset): New function.
3072 * function.c (locate_and_pad_parm): Use
3073 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
3074 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
3075 (TARGET_FUNCTION_ARG_OFFSET): ...this.
3076 * doc/tm.texi: Regenerate.
3077 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
3078 * config/spu/spu.c (spu_function_arg_offset): New function.
3079 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
3080 * system.h (FUNCTION_ARG_OFFSET): Poison.
3082 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
3083 Alan Hayard <alan.hayward@arm.com>
3084 David Sherwood <david.sherwood@arm.com>
3086 * target.def (truly_noop_truncation): New hook.
3087 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
3088 than TRULY_NOOP_TRUNCATION.
3089 * hooks.h (hook_bool_uint_uint_true): Declare.
3090 * hooks.c (hook_bool_uint_uint_true): New function.
3091 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
3092 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
3093 * doc/tm.texi: Regenerate.
3094 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
3095 rather than TRULY_NOOP_TRUNCATION in comments.
3096 (simplify_comparison): Likewise.
3097 (record_truncated_value): Likewise.
3098 * expmed.c (extract_bit_field_1): Likewise.
3099 (extract_split_bit_field): Likewise.
3100 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
3101 instead of TRULY_NOOP_TRUNCATION.
3102 * function.c (assign_parm_setup_block): Likewise.
3103 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
3104 * rtlhooks.c: Include target.h.
3105 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
3106 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
3107 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
3108 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
3109 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
3110 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
3111 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
3112 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
3113 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
3114 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
3115 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
3116 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
3117 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
3118 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
3119 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
3120 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
3121 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
3122 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
3123 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
3124 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
3125 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
3126 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
3127 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
3128 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
3129 * config/mips/mips.c (mips_truly_noop_truncation): New function.
3130 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
3131 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
3132 rather than TRULY_NOOP_TRUNCATION in comments.
3133 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
3134 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
3135 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
3136 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
3137 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
3138 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
3139 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
3140 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
3141 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
3142 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
3143 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
3144 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
3145 rather than TRULY_NOOP_TRUNCATION in comments.
3146 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
3147 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
3148 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
3149 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
3150 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
3151 TRULY_NOOP_TRUNCATION condition.
3152 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
3153 (TRULY_NOOP_TRUNCATION): Delete.
3154 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
3155 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
3156 * config/spu/spu.c (spu_truly_noop_truncation): New function.
3157 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
3158 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
3159 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
3160 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
3161 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
3162 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
3163 rather than TRULY_NOOP_TRUNCATION in comments.
3164 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
3165 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
3166 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
3167 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
3168 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
3169 * system.h (TRULY_NOOP_TRUNCATION): Poison.
3171 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
3174 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
3175 (*cmp_ior): Likewise.
3176 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
3177 (*ior_scc_scc_cmp): Likewise.
3178 (*and_scc_scc): Likewise.
3179 (*and_scc_scc_cmp): Likewise.
3181 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
3182 Alan Hayard <alan.hayward@arm.com>
3183 David Sherwood <david.sherwood@arm.com>
3185 * target.def (can_change_mode_class): New hook.
3186 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
3187 (hard_regno_nregs): Likewise.
3188 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
3189 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
3190 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
3191 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
3192 (LOAD_EXTEND_OP): Update accordingly.
3193 * doc/tm.texi: Regenerate.
3194 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
3195 CANNOT_CHANGE_MODE_CLASS.
3196 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
3197 (REG_CAN_CHANGE_MODE_P): ...this new macro.
3198 * combine.c (simplify_set): Update accordingly.
3199 * emit-rtl.c (validate_subreg): Likewise.
3200 * recog.c (general_operand): Likewise.
3201 * regcprop.c (mode_change_ok): Likewise.
3202 * reload1.c (choose_reload_regs): Likewise.
3203 (inherit_piecemeal_p): Likewise.
3204 * rtlanal.c (simplify_subreg_regno): Likewise.
3205 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
3206 instead of CANNOT_CHANGE_MODE_CLASS.
3207 (reload_cse_simplify_operands): Likewise.
3208 * reload.c (push_reload): Use targetm.can_change_mode_class
3209 instead of CANNOT_CHANGE_MODE_CLASS.
3210 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
3211 REG_CANNOT_CHANGE_MODE_P.
3212 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3213 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
3214 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3215 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3216 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3217 (arm_can_change_mode_class): New function.
3218 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
3219 than CANNOT_CHANGE_MODE_CLASS in comments.
3220 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3221 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
3222 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
3223 (ix86_can_change_mode_class): ...this new function, inverting the
3224 sense of the return value.
3225 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3226 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3227 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3228 (ia64_can_change_mode_class): New function.
3229 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3230 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
3231 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
3232 (m32c_can_change_mode_class): ...this new function, inverting the
3233 sense of the return value.
3234 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3235 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3236 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
3237 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
3238 (mips_can_change_mode_class): ...this new function, inverting the
3239 sense of the return value.
3240 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3241 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3242 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3243 (msp430_can_change_mode_class): New function.
3244 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3245 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
3246 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3247 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3248 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3249 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
3250 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3251 (pa_cannot_change_mode_class): Replace with...
3252 (pa_can_change_mode_class): ...this new function, inverting the
3253 sense of the return value.
3254 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
3255 than CANNOT_CHANGE_MODE_CLASS in comments.
3256 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3257 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
3258 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3259 (pdp11_cannot_change_mode_class): Replace with...
3260 (pdp11_can_change_mode_class): ...this new function, inverting the
3261 sense of the return value.
3262 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3263 * config/powerpcspe/powerpcspe-protos.h
3264 (rs6000_cannot_change_mode_class_ptr): Delete.
3265 * config/powerpcspe/powerpcspe.c
3266 (rs6000_cannot_change_mode_class_ptr): Delete.
3267 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3268 (rs6000_option_override_internal): Assign to
3269 targetm.can_change_mode_class instead of
3270 rs6000_cannot_change_mode_class_ptr.
3271 (rs6000_cannot_change_mode_class): Replace with...
3272 (rs6000_can_change_mode_class): ...this new function, inverting the
3273 sense of the return value.
3274 (rs6000_debug_cannot_change_mode_class): Replace with...
3275 (rs6000_debug_can_change_mode_class): ...this new function.
3276 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3277 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
3278 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3279 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3280 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
3282 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
3283 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3284 (rs6000_option_override_internal): Assign to
3285 targetm.can_change_mode_class instead of
3286 rs6000_cannot_change_mode_class_ptr.
3287 (rs6000_cannot_change_mode_class): Replace with...
3288 (rs6000_can_change_mode_class): ...this new function, inverting the
3289 sense of the return value.
3290 (rs6000_debug_cannot_change_mode_class): Replace with...
3291 (rs6000_debug_can_change_mode_class): ...this new function.
3292 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3293 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
3294 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
3295 (s390_can_change_mode_class): ...this new function, inverting the
3296 sense of the return value.
3297 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3298 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3299 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
3300 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3301 (sh_cannot_change_mode_class): Replace with...
3302 (sh_can_change_mode_class): ...this new function, inverting the
3303 sense of the return value.
3304 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3305 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3306 (sparc_can_change_mode_class): New function.
3307 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3308 * config/spu/spu.c (spu_can_change_mode_class): New function.
3309 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3310 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3311 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3312 (visium_can_change_mode_class): New function.
3313 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
3315 2017-09-15 Richard Biener <rguenther@suse.de>
3317 PR tree-optimization/82217
3318 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
3319 but not undefined case.
3321 2017-09-15 Jakub Jelinek <jakub@redhat.com>
3324 * postreload.c (reload_cse_simplify_operands): Skip
3325 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
3327 2017-09-15 Richard Biener <rguenther@suse.de>
3329 PR tree-optimization/68823
3330 * graphite-scop-detection.c (build_alias_set): If we have a
3331 possible dependence check whether we can handle them by just
3332 looking at the DRs DR_ACCESS_FNs.
3333 (build_scops): If build_alias_set fails, fail the SCOP.
3335 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
3337 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
3338 to support float128 built-in functions that require the ISA 3.0
3340 (BU_FLOAT128_3_HW): Likewise.
3341 (SQRTF128): Add support for the IEEE 128-bit square root and fma
3343 (FMAF128): Likewise.
3345 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
3346 support for built-in functions that need the ISA 3.0 IEEE 128-bit
3347 floating point instructions.
3348 (rs6000_invalid_builtin): Likewise.
3349 (rs6000_builtin_mask_names): Likewise.
3350 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
3351 (RS6000_BTM_FLOAT128_HW): Likewise.
3352 (RS6000_BTM_COMMON): Likewise.
3353 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
3355 * doc/extend.texi (RS/6000 built-in functions): Document the
3356 IEEE 128-bit floating point square root and fused multiply-add
3359 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
3361 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
3362 reg (r2) isn't in the set of registers defined in the prologue.
3364 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3365 Alan Hayward <alan.hayward@arm.com>
3366 David Sherwood <david.sherwood@arm.com>
3368 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
3369 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
3370 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
3371 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
3372 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
3374 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
3375 max_vectorization_factor.
3376 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
3378 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3379 Alan Hayward <alan.hayward@arm.com>
3380 David Sherwood <david.sherwood@arm.com>
3382 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
3383 (vect_worthwhile_without_simd_p): Declare.
3384 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
3385 (vectorizable_reduction): Use it.
3386 * tree-vect-stmts.c (vectorizable_shift): Likewise.
3387 (vectorizable_operation): Likewise.
3389 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3390 Alan Hayward <alan.hayward@arm.com>
3391 David Sherwood <david.sherwood@arm.com>
3393 * tree-vectorizer.h (vect_get_num_copies): New function.
3394 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
3395 * tree-vect-loop.c (vectorizable_reduction): Likewise.
3396 (vectorizable_induction): Likewise.
3397 (vectorizable_live_operation): Likewise.
3398 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
3399 (vectorizable_bswap): Likewise.
3400 (vectorizable_call): Likewise.
3401 (vectorizable_conversion): Likewise.
3402 (vectorizable_assignment): Likewise.
3403 (vectorizable_shift): Likewise.
3404 (vectorizable_operation): Likewise.
3405 (vectorizable_store): Likewise.
3406 (vectorizable_load): Likewise.
3407 (vectorizable_condition): Likewise.
3408 (vectorizable_comparison): Likewise.
3409 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
3411 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3412 Alan Hayward <alan.hayward@arm.com>
3413 David Sherwood <david.sherwood@arm.com>
3415 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
3416 of vect_init_vector.
3418 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3419 Alan Hayward <alan.hayward@arm.com>
3420 David Sherwood <david.sherwood@arm.com>
3422 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
3423 an inline wrapper that provides a location.
3424 (gimple_build_vector): Likewise.
3425 * gimple-fold.c (gimple_build_vector_from_val): New function.
3426 (gimple_build_vector): Likewise.
3427 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
3428 functions to build the initial value. Always return a gimple value.
3429 (get_initial_defs_for_reduction): Likewise. Only compute
3431 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
3432 vect_init_vector on the results from get_initial_def(s)_for_reduction.
3433 (vectorizable_induction): Use gimple_build_vector rather than
3436 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3437 Alan Hayward <alan.hayward@arm.com>
3438 David Sherwood <david.sherwood@arm.com>
3440 * target.h (vec_perm_indices): New typedef.
3441 (auto_vec_perm_indices): Likewise.
3442 * optabs-query.h: Include target.h
3443 (can_vec_perm_p): Take a vec_perm_indices *.
3444 * optabs-query.c (can_vec_perm_p): Likewise.
3445 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
3446 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
3447 * tree-vect-generic.c (lower_vec_perm): Likewise.
3448 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
3449 (vect_grouped_load_supported): Likewise.
3450 (vect_shift_permute_load_chain): Likewise.
3451 (vect_permute_store_chain): Use auto_vec_perm_indices.
3452 (vect_permute_load_chain): Likewise.
3453 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
3454 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
3455 Update uses of can_vec_perm_p.
3456 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
3457 mode with a number of elements. Take a vec_perm_indices *.
3458 (vect_create_epilog_for_reduction): Update accordingly.
3459 Use auto_vec_perm_indices.
3460 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
3461 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
3462 (vect_transform_slp_perm_load): Likewise.
3463 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
3464 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
3465 (vect_gen_perm_mask_checked): Likewise.
3466 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
3467 (vect_gen_perm_mask_checked): Likewise.
3468 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
3469 (vectorizable_store): Likewise.
3470 (vectorizable_load): Likewise.
3471 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
3472 (vectorizable_bswap): Likewise.
3474 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3475 Alan Hayward <alan.hayward@arm.com>
3476 David Sherwood <david.sherwood@arm.com>
3478 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
3479 * tree.c (build_vector): Likewise.
3480 (build_vector_from_ctor): Update accordingly.
3481 (build_vector_from_val): Likewise.
3482 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
3483 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
3484 * tree-vect-generic.c (add_rshift): Likewise.
3485 (expand_vector_divmod): Likewise.
3486 (optimize_vector_constructor): Likewise.
3487 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
3488 (vect_transform_slp_perm_load): Likewise.
3489 (vect_schedule_slp_instance): Likewise.
3490 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
3491 (vectorizable_call): Likewise.
3492 (vect_gen_perm_mask_any): Likewise. Add elements in order.
3493 * expmed.c (make_tree): Likewise.
3494 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
3495 a vector passed to build_vector.
3496 (fold_convert_const): Likewise.
3497 (exact_inverse): Likewise.
3498 (fold_ternary_loc): Likewise.
3499 (fold_relational_const): Likewise.
3500 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
3501 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
3502 (const_unop): Likewise. Store the reduction accumulator in a
3503 variable rather than an array.
3504 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
3505 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
3506 the new vector, rather than constructing it after the input arrays.
3507 (native_interpret_vector): Use auto_vec<tree> when building
3508 a vector passed to build_vector. Add elements in order.
3509 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
3510 auto_vec<tree> when building a vector passed to build_vector.
3511 (vect_create_epilog_for_reduction): Likewise.
3512 (vectorizable_induction): Likewise.
3513 (get_initial_def_for_reduction): Likewise. Fix indentation of
3515 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
3517 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
3518 passed to build_vector.
3520 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3521 Alan Hayward <alan.hayward@arm.com>
3522 David Sherwood <david.sherwood@arm.com>
3524 * tree-core.h (tree_base::u): Add an "nelts" field.
3525 (tree_vector): Use VECTOR_CST_NELTS as the length.
3526 * tree.c (tree_size): Likewise.
3527 (make_vector): Initialize VECTOR_CST_NELTS.
3528 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
3529 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
3530 TYPE_VECTOR_SUBPARTS.
3531 * expr.c (const_vector_mask_from_tree): Consistently use "units"
3532 as the number of units, setting it from VECTOR_CST_NELTS.
3533 (const_vector_from_tree): Likewise.
3534 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
3535 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
3536 (fold_negate_expr_1): Likewise.
3537 (fold_convert_const): Likewise.
3538 (const_binop): Likewise. Differentiate the number of output and
3540 (const_unop): Likewise.
3541 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
3542 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
3543 in cases that did the opposite.
3545 2017-09-14 Richard Biener <rguenther@suse.de>
3547 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
3550 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
3552 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
3554 2017-09-14 Jakub Jelinek <jakub@redhat.com>
3557 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
3558 if and where to split a bb, except for splitting before debug insn
3559 sequences followed by non-label real insn. Delete debug insns
3560 in between basic blocks.
3562 * combine.c (make_compound_operation_int): Formatting fixes.
3564 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
3565 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3566 * config/netbsd.h (LINK_EH_SPEC): Likewise.
3567 * config/sol2.h (LINK_EH_SPEC): Likewise.
3568 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3569 * config/s390/linux.h (LINK_SPEC): Likewise.
3570 * config/freebsd.h (LINK_EH_SPEC): Likewise.
3571 * config/openbsd.h (LINK_EH_SPEC): Likewise.
3572 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3573 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
3574 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
3575 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3576 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
3577 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
3579 2017-09-13 Jakub Jelinek <jakub@redhat.com>
3581 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
3583 (ENDFILE_LINUX_SPEC): Likewise.
3584 (LINK_EH_SPEC): Likewise.
3585 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
3586 (LINK_OS_LINUX_SPEC32): Likewise.
3587 (LINK_OS_LINUX_SPEC64): Likewise.
3588 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
3589 (LINK_OS_LINUX_SPEC): Likewise.
3591 2017-09-13 Martin Liska <mliska@suse.cz>
3594 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
3595 CASE_HIGH is NULL_TREE.
3597 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
3598 Alan Hayward <alan.hayward@arm.com>
3599 David Sherwood <david.sherwood@arm.com>
3601 * target.def (secondary_memory_needed): New hook.
3602 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
3603 instead of SECONDARY_MEMORY_NEEDED.
3604 (secondary_memory_needed_mode): Likewise.
3605 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
3606 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
3607 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
3608 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
3609 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
3610 * doc/tm.texi: Regenerate.
3611 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
3612 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
3613 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3614 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
3615 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
3616 * config/i386/i386.c (inline_secondary_memory_needed): Put the
3617 mode argument first and change the reg_class arguments to reg_class_t.
3618 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
3619 Make static. Update the call to inline_secondary_memory_needed.
3620 (ix86_register_move_cost): Update the call to
3621 inline_secondary_memory_needed.
3622 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3623 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
3625 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
3626 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
3628 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
3629 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
3630 * config/mips/mips.c (mips_secondary_memory_needed): Make static
3631 and match hook interface. Add comment from mips.h.
3632 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3633 * config/mmix/mmix.md (truncdfsf2): Refer to
3634 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
3636 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
3637 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
3638 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3639 (pa_secondary_memory_needed): New function.
3640 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
3641 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
3642 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3643 (pdp11_secondary_memory_needed): Make static and match hook interface.
3644 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
3645 * config/powerpcspe/powerpcspe-protos.h
3646 (rs6000_secondary_memory_needed_ptr): Delete.
3647 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
3649 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3650 (rs6000_option_override_internal): Assign to
3651 targetm.secondary_memory_needed rather than
3652 rs6000_secondary_memory_needed_ptr.
3653 (rs6000_secondary_memory_needed): Match hook interface.
3654 (rs6000_debug_secondary_memory_needed): Likewise.
3655 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
3656 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
3657 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
3658 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3659 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
3660 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
3662 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
3663 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3664 (rs6000_option_override_internal): Assign to
3665 targetm.secondary_memory_needed rather than
3666 rs6000_secondary_memory_needed_ptr.
3667 (rs6000_secondary_memory_needed): Match hook interface.
3668 (rs6000_debug_secondary_memory_needed): Likewise.
3669 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
3670 * config/s390/s390.c (s390_secondary_memory_needed): New function.
3671 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3672 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
3673 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3674 (sparc_secondary_memory_needed): New function.
3675 * lra-constraints.c (check_and_process_move): Refer to
3676 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
3678 (curr_insn_transform): Likewise.
3679 (process_alt_operands): Use targetm.secondary_memory_needed
3680 instead of TARGET_SECONDARY_MEMORY_NEEDED.
3681 (check_secondary_memory_needed_p): Likewise.
3682 (choose_split_class): Likewise.
3683 * reload.c: Unconditionally include code that was previously
3684 conditional on SECONDARY_MEMORY_NEEDED.
3685 (push_secondary_reload): Use targetm.secondary_memory_needed
3686 instead of TARGET_SECONDARY_MEMORY_NEEDED.
3687 (push_reload): Likewise.
3688 * reload1.c: Unconditionally include code that was previously
3689 conditional on SECONDARY_MEMORY_NEEDED.
3690 (choose_reload_regs): Use targetm.secondary_memory_needed
3691 instead of TARGET_SECONDARY_MEMORY_NEEDED.
3692 (gen_reload): Likewise.
3693 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
3695 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
3696 Alan Hayward <alan.hayward@arm.com>
3697 David Sherwood <david.sherwood@arm.com>
3699 * target.def (secondary_memory_needed_mode): New hook:
3700 * targhooks.c (default_secondary_memory_needed_mode): Declare.
3701 * targhooks.h (default_secondary_memory_needed_mode): New function.
3702 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
3703 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
3704 * doc/tm.texi: Regenerate.
3705 * lra-constraints.c (check_and_process_move): Use
3706 targetm.secondary_memory_needed_mode instead of
3707 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
3708 (curr_insn_transform): Likewise.
3709 * reload.c (get_secondary_mem): Likewise.
3710 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3711 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
3713 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3714 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3715 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
3716 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3717 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
3719 * config/powerpcspe/powerpcspe-protos.h
3720 (rs6000_secondary_memory_needed_mode): Delete.
3721 * config/powerpcspe/powerpcspe.c
3722 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3723 (rs6000_secondary_memory_needed_mode): Make static.
3724 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3725 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
3727 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
3729 (rs6000_secondary_memory_needed_mode): Make static.
3730 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3731 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
3732 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3733 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3734 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
3736 (sparc_secondary_memory_needed_mode): New function.
3737 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
3739 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
3741 * config/aarch64/constraints.md (Umq): New constraint.
3742 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
3744 (mov<mode>): Update condition.
3746 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3748 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
3749 when bitposition is the same.
3751 2017-09-13 Richard Biener <rguenther@suse.de>
3753 * dwarf2out.c (output_die_symbol): Remove.
3754 (output_die): Do not output a DIEs symbol.
3756 2017-09-13 Richard Biener <rguenther@suse.de>
3759 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
3760 default-def to avoid breaking iterator update with the weird
3761 interaction with cgraph_update_edges_for_call_stmt_node.
3763 2017-09-13 Richard Biener <rguenther@suse.de>
3765 * tree-cfg.c (verify_gimple_assign_binary): Add verification
3766 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
3767 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
3768 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
3770 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
3772 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
3773 Disable pc relative literal load irrespective of
3774 TARGET_FIX_ERR_A53_84341 for default.
3776 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
3778 * config/sparc/sparc.c (output_return): Output the source location of
3779 the insn in the delay slot, if any.
3780 (output_sibcall): Likewise.
3782 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
3785 * common.opt (-static-pie): New alias.
3786 (shared): Negate static-pie.
3787 (-no-pie): Update help text.
3789 (static-pie): New option.
3790 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
3791 -static-pie support.
3792 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
3793 (LINK_EH_SPEC): Likewise.
3794 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3795 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
3796 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
3797 * gcc.c (LINK_COMMAND_SPEC): Likewise.
3798 (init_gcc_specs): Likewise.
3799 (init_spec): Likewise.
3800 (display_help): Update help message for -pie.
3801 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
3804 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
3806 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
3807 (movdi_aarch64): Likewise.
3808 (movti_aarch64): Likewise.
3810 2017-09-12 Simon Wright <simon@pushface.org>
3813 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
3814 calculation of the minor version, always output as 0.
3816 2017-09-12 Jakub Jelinek <jakub@redhat.com>
3819 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
3820 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
3821 on it early, rather than manual conversion late. For
3822 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
3823 instead of performing manual conversion.
3825 2017-09-12 Carl Love <cel@us.ibm.com>
3827 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
3828 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
3830 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
3831 VMULOSW): Add definitions.
3832 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
3833 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
3834 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
3835 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
3836 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
3838 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
3840 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
3842 (movti_aarch64): Likewise.
3843 (movdf_aarch64): Likewise.
3844 (movtf_aarch64): Likewise.
3845 (load_pairdi): Likewise.
3846 (store_pairdi): Likewise.
3847 (load_pairdf): Likewise.
3848 (store_pairdf): Likewise.
3849 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
3850 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
3851 (ldr_got_small_<mode>): Likewise.
3852 (ldr_got_small_28k_<mode>): Likewise.
3853 (ldr_got_tiny): Likewise.
3854 * config/aarch64/iterators.md (ldst_sz): New.
3855 (ldpstp_sz): Likewise.
3856 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
3858 (thunderx_load): Split load_8 to load_16.
3859 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
3861 (thunderx2t99_storepair_basic): Split store_8 to store_16.
3862 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
3863 (xgene1_store_pair): Split store_8 to store_16.
3864 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
3865 (falkor_st_0_st_sd): Split store_8 to store_16.
3867 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
3869 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
3870 and store1/2/3/4 to store_4/8/12/16.
3871 * config/aarch64/aarch64.md: Update for rename.
3872 * config/arm/arm.md: Likewise.: Likewise.
3873 * config/arm/arm.c: Likewise.
3874 * config/arm/thumb1.md: Likewise.
3875 * config/arm/thumb2.md: Likewise.
3876 * config/arm/vfp.md: Likewise.
3877 * config/arm/arm-generic.md: Likewise.
3878 * config/arm/arm1020e.md: Likewise.
3879 * config/arm/arm1026ejs.md: Likewise.
3880 * config/arm/arm1136jfs.md: Likewise.
3881 * config/arm/arm926ejs.md: Likewise.
3882 * config/arm/cortex-a15.md: Likewise.
3883 * config/arm/cortex-a17.md: Likewise.
3884 * config/arm/cortex-a5.md: Likewise.
3885 * config/arm/cortex-a53.md: Likewise.
3886 * config/arm/cortex-a57.md: Likewise.
3887 * config/arm/cortex-a7.md: Likewise.
3888 * config/arm/cortex-a8.md: Likewise.
3889 * config/arm/cortex-a9.md: Likewise.
3890 * config/arm/cortex-m4.md: Likewise.
3891 * config/arm/cortex-m7.md: Likewise.
3892 * config/arm/cortex-r4.md: Likewise.
3893 * config/arm/exynos-m1.md: Likewise.
3894 * config/arm/fa526.md: Likewise.
3895 * config/arm/fa606te.md: Likewise.
3896 * config/arm/fa626te.md: Likewise.
3897 * config/arm/fa726te.md: Likewise.
3898 * config/arm/fmp626.md: Likewise.
3899 * config/arm/iwmmxt.md: Likewise.
3900 * config/arm/ldmstm.md: Likewise.
3901 * config/arm/marvell-pj4.md: Likewise.
3902 * config/arm/xgene1.md: Likewise.
3903 * config/aarch64/thunderx.md: Likewise.
3904 * config/aarch64/thunderx2t99.md: Likewise.
3905 * config/aarch64/falkor.md: Likewise.
3907 2017-09-12 Martin Liska <mliska@suse.cz>
3909 * attribs.c (private_lookup_attribute): New function.
3910 * attribs.h (private_lookup_attribute): Declared here.
3911 (lookup_attribute): Called from this place.
3913 2017-09-12 Richard Biener <rguenther@suse.de>
3915 PR tree-optimization/82157
3916 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
3917 stmts with side-effects.
3919 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3920 Alan Hayward <alan.hayward@arm.com>
3921 David Sherwood <david.sherwood@arm.com>
3923 * target.def (hard_regno_nregs): New hook.
3924 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
3925 * targhooks.h (default_hard_regno_nregs): Declare.
3926 * targhooks.c (default_hard_regno_nregs): New function.
3927 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
3928 (TARGET_HARD_REGNO_NREGS): ...this hook.
3929 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
3930 (CLASS_MAX_NREGS): Likewise.
3931 * doc/tm.texi: Regenerate.
3932 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
3933 instead of HARD_REGNO_NREGS.
3934 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
3935 HARD_REGNO_NREGS in the comment.
3936 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
3937 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
3938 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
3939 Return an unsigned int.
3940 (TARGET_HARD_REGNO_NREGS): Redefine.
3941 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
3942 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
3943 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
3944 (arc_hard_regno_nregs): New function.
3945 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
3946 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
3947 (arm_hard_regno_nregs): New function.
3948 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
3949 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
3950 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
3951 (TARGET_HARD_REGNO_NREGS): Redefine.
3952 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
3953 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
3954 (HARD_REGNO_NREGS): Delete.
3955 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
3956 (cr16_hard_regno_nregs): New function.
3957 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
3958 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
3959 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
3960 (cris_hard_regno_nregs): New function.
3961 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
3962 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
3963 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
3964 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
3965 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
3966 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
3967 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
3968 (frv_hard_regno_nregs): Make static. Take and return an
3970 (frv_class_max_nregs): Remove outdated copy of documentation.
3971 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
3972 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
3973 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
3974 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
3975 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
3976 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
3977 (TARGET_HARD_REGNO_NREGS): Redefine.
3978 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
3979 (CLASS_MAX_NREGS): Update comment.
3980 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
3981 (ia64_hard_regno_nregs): New function.
3982 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
3983 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
3984 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
3985 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
3986 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
3988 (m32c_hard_regno_nregs): Likewise. Make static.
3989 (TARGET_HARD_REGNO_NREGS): Redefine.
3990 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
3991 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
3992 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
3993 (m68k_hard_regno_nregs): New function.
3994 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
3995 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
3996 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
3997 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
3998 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
3999 Take and return an unsigned int.
4000 (TARGET_HARD_REGNO_NREGS): Redefine.
4001 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
4002 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
4003 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
4004 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
4005 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
4006 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
4007 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
4008 (msp430_hard_regno_nregs): Make static. Take and return an
4010 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
4011 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
4012 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
4013 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
4014 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
4015 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
4016 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
4017 (TARGET_HARD_REGNO_NREGS): Redefine.
4018 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
4019 (PA_HARD_REGNO_NREGS): ...this.
4020 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
4021 (PA_HARD_REGNO_NREGS): ...this.
4022 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
4023 (pa_hard_regno_nregs): New function.
4024 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
4025 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
4026 (pdp11_hard_regno_nregs): New function.
4027 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
4028 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
4029 (rs6000_hard_regno_nregs_hook): New function.
4030 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
4031 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
4032 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
4033 Take and return an unsigned int. Move earlier in file.
4034 (TARGET_HARD_REGNO_NREGS): Redefine.
4035 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
4036 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
4037 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
4038 (rl78_hard_regno_nregs): Make static. Take and return an
4040 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
4041 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
4042 (rs6000_hard_regno_nregs_hook): New function.
4043 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
4044 * config/rx/rx.c (rx_hard_regno_nregs): New function.
4045 (TARGET_HARD_REGNO_NREGS): Redefine.
4046 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
4047 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
4048 instead of HARD_REGNO_NREGS.
4049 (s390_hard_regno_nregs): New function.
4050 (s390_hard_regno_mode_ok): Add comment from s390.h.
4051 (TARGET_HARD_REGNO_NREGS): Redefine.
4052 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
4053 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
4054 (sh_hard_regno_nregs): New function.
4055 (sh_pass_in_reg_p): Use it.
4056 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
4057 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
4058 (sparc_hard_regno_nregs): New function.
4059 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
4060 * config/spu/spu.c (spu_hard_regno_nregs): New function.
4061 (spu_function_arg_advance): Use it, supplying a valid register number.
4062 (TARGET_HARD_REGNO_NREGS): Redefine.
4063 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
4064 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
4065 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
4066 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
4067 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
4068 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
4069 (CLASS_MAX_NREGS): Remove copy of old documentation.
4070 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
4071 (visium_hard_regno_nregs): New function.
4072 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
4073 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
4074 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
4075 xtensa_hard_regno_nregs): New function.
4076 * system.h (HARD_REGNO_NREGS): Poison.
4078 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4080 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
4081 hard_regno_nregs instead of HARD_REGNO_NREGS.
4082 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
4083 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
4084 (c6x_expand_epilogue): Likewise.
4085 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
4086 (frv_read_iacc_argument): Likewise.
4087 * config/sh/sh.c: Include regs.h.
4088 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
4089 (regs_used): Likewise.
4090 (output_stack_adjust): Likewise.
4091 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
4092 * expmed.c: Include regs.h.
4093 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
4094 * ree.c: Include regs.h.
4095 (combine_reaching_defs): Use hard_regno_nregs instead of
4097 (add_removable_extension): Likewise.
4099 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4101 * regs.h (hard_regno_nregs): Turn into a function.
4102 (end_hard_regno): Update accordingly.
4103 * caller-save.c (setup_save_areas): Likewise.
4104 (save_call_clobbered_regs): Likewise.
4105 (replace_reg_with_saved_mem): Likewise.
4106 (insert_restore): Likewise.
4107 (insert_save): Likewise.
4108 * combine.c (can_change_dest_mode): Likewise.
4109 (move_deaths): Likewise.
4110 (distribute_notes): Likewise.
4111 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
4112 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
4113 (rs6000_split_multireg_move): Likewise.
4114 (rs6000_register_move_cost): Likewise.
4115 (rs6000_memory_move_cost): Likewise.
4116 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
4117 (rs6000_split_multireg_move): Likewise.
4118 (rs6000_register_move_cost): Likewise.
4119 (rs6000_memory_move_cost): Likewise.
4120 * cselib.c (cselib_reset_table): Likewise.
4121 (cselib_lookup_1): Likewise.
4122 * emit-rtl.c (set_mode_and_regno): Likewise.
4123 * function.c (aggregate_value_p): Likewise.
4124 * ira-color.c (setup_profitable_hard_regs): Likewise.
4125 (check_hard_reg_p): Likewise.
4126 (calculate_saved_nregs): Likewise.
4127 (assign_hard_reg): Likewise.
4128 (improve_allocation): Likewise.
4129 (calculate_spill_cost): Likewise.
4130 * ira-emit.c (modify_move_list): Likewise.
4131 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
4132 (ira_hard_reg_in_set_p): Likewise.
4133 * ira.c (setup_reg_mode_hard_regset): Likewise.
4134 (clarify_prohibited_class_mode_regs): Likewise.
4135 (check_allocation): Likewise.
4136 * lra-assigns.c (find_hard_regno_for_1): Likewise.
4137 (lra_setup_reg_renumber): Likewise.
4138 (setup_try_hard_regno_pseudos): Likewise.
4139 (spill_for): Likewise.
4140 (assign_hard_regno): Likewise.
4141 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
4142 * lra-constraints.c (in_class_p): Likewise.
4143 (lra_constraint_offset): Likewise.
4144 (simplify_operand_subreg): Likewise.
4145 (lra_constraints): Likewise.
4146 (split_reg): Likewise.
4147 (split_if_necessary): Likewise.
4148 (invariant_p): Likewise.
4149 (inherit_in_ebb): Likewise.
4150 * lra-lives.c (process_bb_lives): Likewise.
4151 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
4152 (get_hard_regs): Likewise.
4153 (do_remat): Likewise.
4154 * lra-spills.c (assign_spill_hard_regs): Likewise.
4155 * mode-switching.c (create_pre_exit): Likewise.
4156 * postreload.c (reload_combine_recognize_pattern): Likewise.
4157 * recog.c (peep2_find_free_register): Likewise.
4158 * regcprop.c (kill_value_regno): Likewise.
4159 (set_value_regno): Likewise.
4160 (copy_value): Likewise.
4161 (maybe_mode_change): Likewise.
4162 (find_oldest_value_reg): Likewise.
4163 (copyprop_hardreg_forward_1): Likewise.
4164 * regrename.c (check_new_reg_p): Likewise.
4165 (regrename_do_replace): Likewise.
4166 * reload.c (push_reload): Likewise.
4167 (combine_reloads): Likewise.
4168 (find_dummy_reload): Likewise.
4169 (operands_match_p): Likewise.
4170 (find_reloads): Likewise.
4171 (find_equiv_reg): Likewise.
4172 (reload_adjust_reg_for_mode): Likewise.
4173 * reload1.c (count_pseudo): Likewise.
4174 (count_spilled_pseudo): Likewise.
4175 (find_reg): Likewise.
4176 (clear_reload_reg_in_use): Likewise.
4177 (free_for_value_p): Likewise.
4178 (allocate_reload_reg): Likewise.
4179 (choose_reload_regs): Likewise.
4180 (reload_adjust_reg_for_temp): Likewise.
4181 (emit_reload_insns): Likewise.
4182 (delete_output_reload): Likewise.
4183 * rtlanal.c (subreg_get_info): Likewise.
4184 * sched-deps.c (sched_analyze_reg): Likewise.
4185 * sel-sched.c (init_regs_for_mode): Likewise.
4186 (mark_unavailable_hard_regs): Likewise.
4187 (choose_best_reg_1): Likewise.
4188 (verify_target_availability): Likewise.
4189 * valtrack.c (dead_debug_insert_temp): Likewise.
4190 * var-tracking.c (track_loc_p): Likewise.
4191 (emit_note_insn_var_location): Likewise.
4192 * varasm.c (make_decl_rtl): Likewise.
4193 * reginfo.c (choose_hard_reg_mode): Likewise.
4194 (init_reg_modes_target): Refer directly to
4195 this_target_regs->x_hard_regno_nregs.
4197 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4199 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
4200 instead of hard_regno_nregs.
4202 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4204 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
4205 end_hard_regno instead of hard_regno_nregs.
4206 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
4207 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
4208 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
4209 * ira-color.c (improve_allocation): Likewise.
4210 * lra-assigns.c (find_hard_regno_for_1): Likewise.
4211 * lra-lives.c (mark_regno_live): Likewise.
4212 (mark_regno_dead): Likewise.
4213 * lra-remat.c (operand_to_remat): Likewise.
4214 * lra.c (collect_non_operand_hard_regs): Likewise.
4215 * postreload.c (reload_combine_note_store): Likewise.
4216 (move2add_valid_value_p): Likewise.
4217 * reload.c (regno_clobbered_p): Likewise.
4219 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4221 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
4223 * config/v850/v850.c (v850_reorg): Likewise.
4224 * reload.c (refers_to_regno_for_reload_p): Likewise.
4225 (find_equiv_reg): Likewise.
4226 * reload1.c (reload_reg_reaches_end_p): Likewise.
4228 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4230 * caller-save.c (add_used_regs): Use REG_NREGS instead of
4232 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
4233 * config/arm/arm.c (output_move_neon): Likewise.
4234 (arm_attr_length_move_neon): Likewise.
4235 (neon_split_vcombine): Likewise.
4236 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
4237 (c6x_mark_reg_written): Likewise.
4238 (c6x_dwarf_register_span): Likewise.
4239 * config/i386/i386.c (ix86_save_reg): Likewise.
4240 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
4241 (rws_access_reg): Likewise.
4242 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
4243 * mode-switching.c (create_pre_exit): Likewise.
4244 * ree.c (combine_reaching_defs): Likewise.
4245 (add_removable_extension): Likewise.
4246 * regcprop.c (find_oldest_value_reg): Likewise.
4247 (copyprop_hardreg_forward_1): Likewise.
4248 * reload.c (reload_inner_reg_of_subreg): Likewise.
4249 (push_reload): Likewise.
4250 (combine_reloads): Likewise.
4251 (find_dummy_reload): Likewise.
4252 (reload_adjust_reg_for_mode): Likewise.
4253 * reload1.c (find_reload_regs): Likewise.
4254 (forget_old_reloads_1): Likewise.
4255 (reload_reg_free_for_value_p): Likewise.
4256 (reload_adjust_reg_for_temp): Likewise.
4257 (emit_reload_insns): Likewise.
4258 (delete_output_reload): Likewise.
4259 * sel-sched.c (choose_best_reg_1): Likewise.
4260 (choose_best_pseudo_reg): Likewise.
4262 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4263 Alan Hayward <alan.hayward@arm.com>
4264 David Sherwood <david.sherwood@arm.com>
4266 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
4267 * target.def (slow_unaligned_access): New hook.
4268 * targhooks.h (default_slow_unaligned_access): Declare.
4269 * targhooks.c (default_slow_unaligned_access): New function.
4270 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
4271 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
4272 * doc/tm.texi: Regenerate.
4273 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
4274 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
4275 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
4277 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
4278 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
4280 (rs6000_slow_unaligned_access): New function.
4281 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
4282 (expand_block_compare): Likewise.
4283 (expand_strn_compare): Likewise.
4284 (rs6000_rtx_costs): Likewise.
4285 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
4286 (riscv_slow_unaligned_access): Likewise.
4287 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
4288 (riscv_slow_unaligned_access_p): ...this and make static.
4289 (riscv_option_override): Update accordingly.
4290 (riscv_slow_unaligned_access): New function.
4291 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
4292 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
4293 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
4294 (rs6000_slow_unaligned_access): New function.
4295 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
4296 (rs6000_rtx_costs): Likewise.
4297 * config/rs6000/rs6000-string.c (expand_block_compare)
4298 (expand_strn_compare): Use targetm.slow_unaligned_access instead
4299 of SLOW_UNALIGNED_ACCESS.
4300 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
4301 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
4302 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
4303 of SLOW_UNALIGNED_ACCESS.
4304 * expmed.c (simple_mem_bitfield_p): Likewise.
4305 * expr.c (alignment_for_piecewise_move): Likewise.
4306 (emit_group_load_1): Likewise.
4307 (emit_group_store): Likewise.
4308 (copy_blkmode_from_reg): Likewise.
4309 (emit_push_insn): Likewise.
4310 (expand_assignment): Likewise.
4311 (store_field): Likewise.
4312 (expand_expr_real_1): Likewise.
4313 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
4314 * lra-constraints.c (simplify_operand_subreg): Likewise.
4315 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
4316 * gimple-ssa-store-merging.c: Likewise in block comment at start
4318 * tree-ssa-strlen.c: Include target.h.
4319 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
4320 of SLOW_UNALIGNED_ACCESS.
4321 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
4323 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4325 PR rtl-optimization/82185
4326 * expmed.c (emit_store_flag_int): Only test tem if it has been
4329 2017-09-12 Richard Biener <rguenther@suse.de>
4332 * match.pd ((FTYPE) N CMP CST): Fix typo.
4334 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
4336 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
4338 (mips_near_type_p): Add 'short_call' attribute as a synonym
4340 * doc/extend.texi (short_call): Document new function attribute.
4342 2017-09-12 Jakub Jelinek <jakub@redhat.com>
4345 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
4346 assertion check that in the condition.
4347 (get_atomic_generic_size): Likewise. Before testing if parameter
4348 has pointer type, if it has array type, call for C++
4349 default_conversion to perform array-to-pointer conversion.
4351 2017-09-12 Richard Biener <rguenther@suse.de>
4353 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
4354 for operations we cannot scalarize.
4356 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
4358 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
4359 vectors heap vectors. Clean up comments.
4360 Make visited_bbs a reference.
4361 (profitable_jump_thread_path): Make GC
4362 vectors heap vectors. Clean up comments.
4364 (convert_and_register_jump_thread_path): Make GC vectors heap
4366 (check_subpath_and_update_thread_path): Same. Clean up comments.
4367 Make visited_bbs a reference.
4368 (handle_phi): Abstract common code to to
4369 register_jump_thread_path_if_profitable.
4370 Rename VAR_BB to DEF_BB.
4372 Make GC vectors heap vectors.
4373 Make visited_bbs a reference.
4374 (handle_assignment): Same.
4375 (register_jump_thread_path_if_profitable): New.
4376 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
4378 Make GC vectors heap vectors. Clean up comments.
4379 Make visited_bbs a reference.
4380 (find_jump_threads_backwards): Make visited_bbs live in the stack.
4381 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
4384 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
4387 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
4388 words of E_DImode object are reachable by xtensa_uimm8x4 access.
4390 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
4392 Revert r251800 and r251799.
4394 2017-09-11 Martin Jambor <mjambor@suse.cz>
4397 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
4398 arguments in advance.
4399 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
4400 use it to find predecessor edges.
4401 (naive_outof_ssa): Collect vector of predecessors.
4403 2017-09-08 Jason Merrill <jason@redhat.com>
4405 PR c++/70029 - ICE with ref-qualifier and -flto
4406 * langhooks.h (struct lang_hooks_for_types): Add
4407 copy_lang_qualifiers.
4408 * attribs.c (build_type_attribute_qual_variant): Use it.
4409 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
4411 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
4412 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
4414 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
4417 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
4418 (*mulsi3_sp64): New instruction.
4419 (mulsi3): New expander.
4421 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
4423 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
4425 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4427 * sancov.c: Include memmodel.h.
4429 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
4432 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
4435 2017-09-07 Carl Love <cel@us.ibm.com>
4437 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
4438 the sldi instruction.
4440 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
4442 * sancov.c: Include tm_p.h.
4444 2017-09-07 Jakub Jelinek <jakub@redhat.com>
4447 * output.h (switch_to_other_text_partition): New declaration.
4448 * varasm.c (switch_to_other_text_partition): New function.
4449 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
4450 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
4451 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
4452 to the other text partition before emitting LCL label and switch back
4453 after emitting the word after it.
4455 2017-09-07 Richard Biener <rguenther@suse.de>
4457 * passes.def (pass_split_crit_edges): Remove instance before PRE.
4458 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
4459 critical edges here, after loop init.
4460 (pass_data_pre): Remove PROP_no_crit_edges flags.
4461 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
4462 for valueization of call args to avoid leaking VN_TOP.
4463 (visit_use): Assert we do not visit default defs.
4464 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
4465 Use error_mark_node to more easily detect leaking VN_TOP.
4466 All default-defs are varying, not VN_TOP. Mark them visited.
4467 (run_scc_vn): Make code match comment.
4469 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
4471 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
4472 OPTION_MASK_FLOAT128_KEYWORD.
4473 (POWERPC_MASKS): Likewise.
4474 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
4475 support for the -mfloat128-type option, and make -mfloat128
4476 default on PowerPC Linux systems. Define or undefine
4477 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
4478 Define __float128 to be __ieee128 if IEEE 128-bit support is
4479 enabled, or undefine it.
4480 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
4481 Delete defining __FLOAT128_TYPE__.
4482 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
4483 -mfloat128-type option and make -mfloat128 default on PowerPC
4485 (TARGET_FLOAT128_TYPE): Likewise.
4486 (-mfloat128-type): Likewise.
4487 * config/rs6000/rs6000.c (rs6000_option_override_internal):
4488 Delete the -mfloat128-type option and make -mfloat128 default on
4489 PowerPC Linux systems. Always use __ieee128 to be the keyword for
4490 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
4491 128-bit floating point is enabled. Change tests from using
4492 -mfloat128-type to -mfloat128.
4493 (rs6000_mangle_type): Use the correct mangling for the __float128
4494 type even if normal long double is restricted to 64-bits.
4495 (floatn_mode): Enable the _Float128 type by default on VSX Linux
4497 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
4498 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
4499 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
4501 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
4502 documentation for -mfloat128.
4504 2017-09-06 Olivier Hainque <hainque@adacore.com>
4506 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
4508 2017-09-06 Wish Wu <wishwu007@gmail.com>
4509 Jakub Jelinek <jakub@redhat.com>
4511 * asan.c (initialize_sanitizer_builtins): Add
4512 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
4513 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
4514 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
4515 BT_FN_VOID_UINT64_PTR variables.
4516 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
4517 (BT_FN_VOID_UINT16_UINT16): Likewise.
4518 (BT_FN_VOID_UINT32_UINT32): Likewise.
4519 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
4520 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
4521 (BT_FN_VOID_UINT64_PTR): Likewise.
4522 * common.opt (flag_sanitize_coverage): New variable.
4523 (fsanitize-coverage=trace-pc): Remove.
4524 (fsanitize-coverage=): Add.
4525 * flag-types.h (enum sanitize_coverage_code): New enum.
4526 * fold-const.c (fold_range_test): Disable non-short-circuit
4527 optimization if flag_sanitize_coverage.
4528 (fold_truth_andor): Likewise.
4529 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
4530 * opts.c (COVERAGE_SANITIZER_OPT): Define.
4531 (coverage_sanitizer_opts): New array.
4532 (get_closest_sanitizer_option): Add OPTS argument, handle also
4533 OPT_fsanitize_coverage_.
4534 (parse_sanitizer_options): Adjusted to also handle
4535 OPT_fsanitize_coverage_.
4536 (common_handle_option): Add OPT_fsanitize_coverage_.
4537 * sancov.c (instrument_comparison, instrument_switch): New function.
4538 (sancov_pass): Add trace-cmp support.
4539 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
4540 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
4541 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
4542 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
4543 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
4544 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
4545 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
4546 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
4547 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
4548 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
4550 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
4552 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
4553 error. Only quit immediately if parsing is complete.
4554 (BEGIN): Initialize fatal_err and parse_done.
4555 (begin fpu, end fpu): Check number of arguments.
4556 (begin arch, end arch): Likewise.
4557 (begin cpu, end cpu): Likewise.
4558 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
4559 (optalias): Likewise.
4561 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
4563 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
4564 * config/arm/arm-isa.h: Delete. Move definitions to ...
4565 * arm-cpus.in: ... here. Use new feature and fgroup values.
4566 * config/arm/arm.c (arm_option_override): Use lower case for feature
4568 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
4569 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
4570 * config/arm/parsecpu.awk (END): Add new command 'isa'.
4572 (print_isa_bits_for): New function.
4573 (gen_isa): New function.
4574 (gen_comm_data): Use print_isa_bits_for.
4575 (define feature): New keyword.
4576 (define fgroup): New keyword.
4577 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
4578 (arm-isa.h): Add rule to generate file.
4579 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
4580 case for feature bit names.
4582 2017-09-06 Richard Biener <rguenther@suse.de>
4584 * tree-ssa-pre.c (NECESSARY): Remove.
4585 (create_expression_by_pieces): Do not touch pass-local flags.
4586 (insert_into_preds_of_block): Likewise.
4587 (do_pre_regular_insertion): Likewise.
4588 (eliminate_insert): Likewise.
4589 (eliminate_dom_walker::before_dom_children): Likewise.
4590 (fini_eliminate): Do not look at inserted_exprs.
4591 (mark_operand_necessary): Remove.
4592 (remove_dead_inserted_code): Replace with simple work-list
4593 algorithm based on inserted_exprs and SSA uses.
4594 (pass_pre::execute): Re-order fini_eliminate and
4595 remove_dead_inserted_code.
4597 2017-09-06 Olivier Hainque <hainque@adacore.com>
4599 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
4600 for VxWorks 7. Adjust surrounding comments.
4602 2017-09-06 Richard Biener <rguenther@suse.de>
4604 * gimple-ssa-strength-reduction.c
4605 (find_candidates_dom_walker::before_dom_children): Also allow
4608 2017-09-06 Richard Biener <rguenther@suse.de>
4610 PR tree-optimization/82108
4611 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
4612 for gap in the non-permutation SLP case.
4614 2017-09-06 Martin Jambor <mjambor@suse.cz>
4616 PR tree-optimization/82078
4617 * tree-sra.c (sort_and_splice_var_accesses): Move call to
4618 add_access_to_work_queue...
4619 (build_accesses_from_assign): ...here.
4620 (propagate_all_subaccesses): Make sure racc is the group
4621 representative, if there is one.
4623 2017-09-06 Jakub Jelinek <jakub@redhat.com>
4626 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
4629 2017-09-06 Richard Biener <rguenther@suse.de>
4631 * gimple-ssa-strength-reduction.c
4632 (find_candidates_dom_walker::before_doom_children): Use a
4633 type and not a mode check.
4635 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
4638 * config/arm/predicates.md (arm_general_adddi_operand): Create new
4640 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
4642 2017-09-05 Jeff Law <law@redhat.com>
4644 PR tree-optimization/64910
4645 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
4646 cases where we have 3 or more operands.
4648 2017-09-05 Jakub Jelinek <jakub@redhat.com>
4651 * omp-low.c (lower_omp_for): Recompute tree invariant if
4652 gimple_omp_for_initial/final is ADDR_EXPR.
4655 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
4656 into gimple val before gimplification fo the COND_EXPR.
4658 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
4660 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
4661 REGION_COPY argument.
4662 (thread_through_all_blocks): Remove unused argument to
4663 duplicate_thread_path.
4665 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4666 Alan Hayward <alan.hayward@arm.com>
4667 David Sherwood <david.sherwood@arm.com>
4669 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
4670 Take a scalar_mode rather than a machine_mode.
4671 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
4672 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
4673 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
4674 (aarch64_gen_adjusted_ldpstp): Likewise.
4675 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
4677 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4678 Alan Hayward <alan.hayward@arm.com>
4679 David Sherwood <david.sherwood@arm.com>
4681 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
4682 Take a scalar_int_mode instead of a machine_mode.
4683 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
4684 (aarch64_output_scalar_simd_mov_immediate): Likewise.
4685 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
4686 (aarch64_simd_attr_length_rglist): Delete.
4687 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
4688 a scalar_int_mode instead of a machine_mode.
4689 (aarch64_add_offset): Likewise.
4690 (aarch64_internal_mov_immediate): Likewise
4691 (aarch64_add_constant_internal): Likewise.
4692 (aarch64_add_constant): Likewise.
4693 (aarch64_movw_imm): Likewise.
4694 (aarch64_rtx_arith_op_extract_p): Likewise.
4695 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
4696 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
4697 Remove assert that the mode isn't a vector.
4698 (aarch64_output_scalar_simd_mov_immediate): Likewise.
4699 (aarch64_expand_mov_immediate): Update calls after above changes.
4700 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
4701 (aarch64_and_bitmask_imm): Check for scalar integer modes.
4702 (aarch64_move_imm): Likewise.
4703 (aarch64_can_const_movi_rtx_p): Likewise.
4704 (aarch64_strip_extend): Likewise.
4705 (aarch64_extr_rtx_p): Likewise.
4706 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
4707 a CONST_INT when the mode parameter is VOIDmode.
4708 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
4710 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4712 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
4713 * stor-layout.c (bitwise_mode_for_mode): Likewise.
4714 (bitwise_type_for_mode): Update accordingly.
4716 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4718 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
4719 * stor-layout.c (mode_for_size_tree): Likewise.
4720 (mode_for_array): Update accordingly.
4721 (layout_decl): Likewise.
4722 (compute_record_mode): Likewise. Only set the mode once.
4724 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4726 * target.def (get_mask_mode): Change return type to opt_mode.
4728 * doc/tm.texi: Regenerate.
4729 * targhooks.h (default_get_mask_mode): Return an opt_mode.
4730 * targhooks.c (default_get_mask_mode): Likewise.
4731 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
4732 * optabs-query.c (can_vec_mask_load_store_p): Update use of
4733 targetm.get_mask_mode.
4734 * tree.c (build_truth_vector_type): Likewise.
4736 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4738 * machmode.h (mode_for_vector): Return an opt_mode.
4739 * stor-layout.c (mode_for_vector): Likewise.
4740 (mode_for_int_vector): Update accordingly.
4741 (layout_type): Likewise.
4742 * config/i386/i386.c (emit_memmov): Likewise.
4743 (ix86_expand_set_or_movmem): Likewise.
4744 (ix86_expand_vector_init): Likewise.
4745 (ix86_get_mask_mode): Likewise.
4746 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
4748 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
4749 * expmed.c (extract_bit_field_1): Likewise.
4750 * expr.c (expand_expr_real_2): Likewise.
4751 * optabs-query.c (can_vec_perm_p): Likewise.
4752 (can_vec_mask_load_store_p): Likewise.
4753 * optabs.c (expand_vec_perm): Likewise.
4754 * targhooks.c (default_get_mask_mode): Likewise.
4755 * tree-vect-stmts.c (vectorizable_store): Likewise.
4756 (vectorizable_load): Likewise.
4757 (get_vectype_for_scalar_type_and_size): Likewise.
4759 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4761 * machmode.h (mode_for_int_vector): New function.
4762 * stor-layout.c (mode_for_int_vector): Likewise.
4763 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
4764 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
4765 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
4766 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
4767 (s390_expand_vcond): Likewise.
4769 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4771 * machmode.h (opt_machine_mode): New type.
4772 (opt_mode<T>): Allow construction from anything that can be
4774 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
4775 (mode_for_size): Return an opt_machine_mode.
4776 * stor-layout.c (mode_for_size): Likewise.
4777 (mode_for_size_tree): Update call accordingly.
4778 (bitwise_mode_for_mode): Likewise.
4779 (make_fract_type): Likewise.
4780 (make_accum_type): Likewise.
4781 * caller-save.c (replace_reg_with_saved_mem): Update call
4783 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4784 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4785 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4786 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4787 * expmed.c (extract_bit_field_1): Likewise.
4788 * reload.c (get_secondary_mem): Likewise.
4789 * varasm.c (assemble_integer): Likewise.
4790 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
4793 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4795 * machmode.h (decimal_float_mode_for_size): New function.
4796 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
4797 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
4798 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
4799 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
4800 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
4801 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
4803 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4805 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
4806 (get_builtin_sync_mode): Likewise.
4807 (expand_ifn_atomic_compare_exchange): Likewise.
4808 (expand_builtin_atomic_clear): Likewise.
4809 (expand_builtin_atomic_test_and_set): Likewise.
4810 (fold_builtin_atomic_always_lock_free): Likewise.
4811 * calls.c (compute_argument_addresses): Likewise.
4812 (emit_library_call_value_1): Likewise.
4813 (store_one_arg): Likewise.
4814 * combine.c (combine_instructions): Likewise.
4815 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
4816 * config/arm/arm.c (arm_function_value): Likewise.
4817 (aapcs_allocate_return_reg): Likewise.
4818 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
4819 * config/i386/i386.c (construct_container): Likewise.
4820 (ix86_gimplify_va_arg): Likewise.
4821 (ix86_expand_sse_cmp): Likewise.
4822 (emit_memmov): Likewise.
4823 (emit_memset): Likewise.
4824 (expand_small_movmem_or_setmem): Likewise.
4825 (ix86_expand_pextr): Likewise.
4826 (ix86_expand_pinsr): Likewise.
4827 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
4828 * config/microblaze/microblaze.c (microblaze_block_move_straight):
4830 * config/mips/mips.c (mips_function_value_1) Likewise.
4831 (mips_block_move_straight): Likewise.
4832 (mips_expand_ins_as_unaligned_store): Likewise.
4833 * config/powerpcspe/powerpcspe.c
4834 (rs6000_darwin64_record_arg_advance_flush): Likewise.
4835 (rs6000_darwin64_record_arg_flush): Likewise.
4836 * config/rs6000/rs6000.c
4837 (rs6000_darwin64_record_arg_advance_flush): Likewise.
4838 (rs6000_darwin64_record_arg_flush): Likewise.
4839 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
4840 (sparc_function_value_1): Likewise.
4841 * config/spu/spu.c (adjust_operand): Likewise.
4842 (spu_emit_branch_or_set): Likewise.
4843 (arith_immediate_p): Likewise.
4844 * emit-rtl.c (gen_lowpart_common): Likewise.
4845 * expr.c (expand_expr_real_1): Likewise.
4846 * function.c (assign_parm_setup_block): Likewise.
4847 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
4848 * reload1.c (alter_reg): Likewise.
4849 * stor-layout.c (mode_for_vector): Likewise.
4850 (layout_type): Likewise.
4852 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4854 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
4855 (spu_convert_move): Likewise.
4856 * lower-subreg.c (resolve_simple_move): Likewise.
4858 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4861 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
4862 define_insn to a define_expand.
4863 (altivec_vsum2sws_direct): New define_insn.
4864 (altivec_vsumsws): Convert from a define_insn to a define_expand.
4866 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
4868 * config/arm/arm.c (arm_option_params_internal): Improve setting of
4871 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
4876 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
4877 realign stack if stack alignment needed is less than incoming
4880 2017-09-05 Marek Polacek <polacek@redhat.com>
4883 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
4886 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
4888 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
4890 2017-09-05 Richard Biener <rguenther@suse.de>
4892 PR tree-optimization/82084
4893 * fold-const.c (can_native_encode_string_p): Handle wide characters.
4895 2017-09-05 Richard Biener <rguenther@suse.de>
4897 PR tree-optimization/82102
4898 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
4900 2017-09-05 Martin Liska <mliska@suse.cz>
4902 PR tree-optimization/82032
4903 * tree-cfg.c (generate_range_test): New function.
4904 * tree-cfg.h (generate_range_test): Declared here.
4905 * tree-cfgcleanup.c (convert_single_case_switch): New function.
4906 (cleanup_control_expr_graph): Use it.
4907 * tree-switch-conversion.c (try_switch_expansion): Remove
4909 (emit_case_nodes): Use generate_range_test.
4911 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
4914 * config/i386/i386.md (*<btsc><mode>_mask): Add
4915 TARGET_USE_BT to insn constraint.
4916 (*btr<mode>_mask): Ditto.
4918 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
4920 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
4921 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
4923 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
4926 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
4927 TARGET_NEON and TARGET_IWMMXT.
4928 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
4929 TARGET_NEON and TARGET_IWMMXT.
4930 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
4932 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
4934 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
4935 (ix86_rewrite_tls_address): Ditto.
4936 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
4937 (ix86_rewrite_tls_address_1): Ditto.
4938 (ix86_rewrite_tls_address): Ditto.
4939 * config/i386/predicates.md (tls_address_pattern): New predicate.
4940 * config/i386/i386.md (TLS address splitter): New splitter.
4942 2017-09-04 Richard Biener <rguenther@suse.de>
4944 PR tree-optimization/82084
4945 * fold-const.h (can_native_encode_string_p): Declare.
4946 * fold-const.c (can_native_encode_string_p): Factor out from ...
4947 (native_encode_string): ... here.
4948 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
4949 vectorizing stores from constants we later cannot handle.
4951 2017-09-04 Marek Polacek <polacek@redhat.com>
4954 * doc/invoke.texi: Update -Wtautological-compare documentation.
4956 2017-09-04 Jeff Law <law@redhat.com>
4958 PR tree-optimization/64910
4959 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
4960 swap the first and last operand if the last is a constant.
4962 2017-09-04 Marek Polacek <polacek@redhat.com>
4965 * convert.c (do_narrow): When sanitizing signed integer overflows,
4966 bail out for signed types.
4967 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
4969 2017-09-04 Richard Biener <rguenther@suse.de>
4971 PR tree-optimization/82060
4972 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
4973 Move devirtualization after stmt folding and before EH/AB/noreturn
4974 cleanup to get the stmt refs canonicalized. Use a bool instead
4975 of gimple_modified_p since that doesn't work for NOPs. Schedule
4976 NOPs generated by folding for removal.
4978 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4979 Alan Hayward <alan.hayward@arm.com>
4980 David Sherwood <david.sherwood@arm.com>
4982 * coretypes.h (pad_direction): New enum.
4983 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
4984 (FUNCTION_ARG_PADDING): Likewise.
4985 * target.def (function_arg_padding): New hook.
4986 * targhooks.h (default_function_arg_padding): Declare.
4987 * targhooks.c (default_function_arg_padding): New function.
4988 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
4989 (TARGET_FUNCTION_ARG_PADDING): ...this.
4990 * doc/tm.texi: Regenerate.
4991 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
4992 instead of direction.
4993 (compute_argument_addresses): Likewise.
4994 (load_register_parameters): Likewise.
4995 (emit_library_call_value_1): Likewise.
4996 (store_one_arg): Use targetm.calls.function_arg_padding instead
4997 of FUNCTION_ARG_PADDING.
4998 (must_pass_in_stack_var_size_or_pad): Likewise.
4999 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
5000 (emit_group_store): Likewise.
5001 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
5002 instead of FUNCTION_ARG_PADDING.
5003 (emit_push_insn): Likewise, and propagate enum change throughout
5005 * function.h (direction): Delete.
5006 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
5008 * function.c (assign_parm_find_stack_rtl): Likewise.
5009 (assign_parm_setup_block_p): Likewise.
5010 (assign_parm_setup_block): Likewise.
5011 (gimplify_parameters): Likewise.
5012 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
5013 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
5015 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
5016 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
5017 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
5018 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
5019 (aarch64_function_arg_padding): ...this new function.
5020 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
5021 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5022 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
5023 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
5024 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
5025 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5026 (arm_pad_arg_upward): Replace with...
5027 (arm_function_arg_padding): ...this new function.
5028 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
5030 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
5031 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
5032 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5033 (ia64_hpux_function_arg_padding): Replace with...
5034 (ia64_function_arg_padding): ...this new function. Use pad_direction
5035 instead of direction. Check for TARGET_HPUX.
5036 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
5037 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5038 (iq2000_function_arg_padding): New function.
5039 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
5040 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
5041 (mips_function_arg_padding): ...this new function.
5042 (mips_pad_reg_upward): Update accordingly.
5043 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5044 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
5045 targetm.calls.function_arg_padding.
5046 (FUNCTION_ARG_PADDING): Delete.
5047 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
5048 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
5049 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
5050 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
5051 (nios2_block_reg_padding): Return pad_direction instead of direction.
5052 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
5053 instead of direction.
5054 (nios2_function_arg_padding): Likewise. Make static.
5055 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5056 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
5057 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
5058 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
5059 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5060 (pa_function_arg_padding): Make static. Return pad_direction instead
5062 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
5063 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
5064 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
5065 instead of direction. Use targetm.calls.function_arg_padding.
5066 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
5067 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
5068 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
5069 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
5070 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
5072 (function_arg_padding): Rename to...
5073 (rs6000_function_arg_padding): ...this. Make static. Return
5074 pad_direction instead of direction.
5075 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
5076 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
5077 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
5078 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
5079 instead of direction. Use targetm.calls.function_arg_padding.
5080 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
5081 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
5082 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
5083 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
5084 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5085 (function_arg_padding): Rename to...
5086 (rs6000_function_arg_padding): ...this. Make static. Return
5087 pad_direction instead of direction.
5088 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
5089 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
5090 * config/s390/s390.c (s390_function_arg_padding): New function.
5091 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5092 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
5093 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
5094 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5095 (function_arg_padding): Rename to...
5096 (sparc_function_arg_padding): ...this. Make static. Return
5097 pad_direction instead of direction.
5098 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
5099 * config/spu/spu.c (spu_function_arg_padding): New function.
5100 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5101 * system.h (FUNCTION_ARG_PADDING): Poison.
5103 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5104 Alan Hayward <alan.hayward@arm.com>
5105 David Sherwood <david.sherwood@arm.com>
5107 * target.def (modes_tieable_p): New hook.
5108 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
5109 (TARGET_MODES_TIEABLE_P): ...this.
5110 * doc/tm.texi.in: Regenerate.
5111 * hooks.h (hook_bool_mode_mode_true): Declare.
5112 * hooks.c (hook_bool_mode_mode_true): New function.
5113 * combine.c (subst): Use targetm.modes_tieable_p instead of
5115 * dse.c (find_shift_sequence): Likewise.
5116 * expmed.c (extract_low_bits): Likewise.
5117 * lower-subreg.c: Include target.h.
5118 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
5120 * rtlanal.c (rtx_cost): Likewise.
5121 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
5122 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
5123 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
5124 (TARGET_MODES_TIEABLE_P): Redefine.
5125 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
5126 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
5127 (TARGET_MODES_TIEABLE_P): Redefine.
5128 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
5129 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
5130 (arc_modes_tieable_p): New function.
5131 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
5132 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
5133 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
5134 (arm_modes_tieable_p): Make static.
5135 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
5136 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
5137 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
5138 (TARGET_MODES_TIEABLE_P): Redefine.
5139 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
5140 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
5141 (TARGET_MODES_TIEABLE_P): Redefine.
5142 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
5143 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
5144 (cr16_modes_tieable_p): New function.
5145 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
5146 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
5147 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
5148 (TRULY_NOOP_TRUNCATION): Update comment.
5149 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
5150 (TRULY_NOOP_TRUNCATION): Update comment.
5151 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
5152 (frv_modes_tieable_p): New function.
5153 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
5154 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
5155 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
5156 (TARGET_MODES_TIEABLE_P): Redefine.
5157 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
5158 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
5159 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
5160 (TARGET_MODES_TIEABLE_P): Redefine.
5161 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
5162 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
5163 (ia64_modes_tieable_p): New function.
5164 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
5165 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
5166 (iq2000_modes_tieable_p): New function.
5167 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
5168 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
5169 (lm32_modes_tieable_p): New function.
5170 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
5171 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
5172 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
5173 (TARGET_MODES_TIEABLE_P): Redefine.
5174 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
5175 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
5176 (m32r_modes_tieable_p): New function.
5177 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
5178 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
5179 (m68k_modes_tieable_p): New function.
5180 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
5181 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
5182 (mcore_modes_tieable_p): New function.
5183 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
5184 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
5186 (TARGET_MODES_TIEABLE_P): Redefine.
5187 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
5188 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
5189 * config/mips/mips.c (mips_modes_tieable_p): Make static.
5190 (TARGET_MODES_TIEABLE_P): Redefine.
5191 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
5192 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
5193 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
5194 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
5195 (mn10300_modes_tieable_p): ...this and make static.
5196 (TARGET_MODES_TIEABLE_P): Redefine.
5197 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
5198 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
5199 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
5200 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
5201 (msp430_modes_tieable_p): Make static.
5202 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
5203 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
5204 (TARGET_MODES_TIEABLE_P): Redefine.
5205 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
5206 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
5207 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
5208 (TARGET_MODES_TIEABLE_P): Redefine.
5209 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
5210 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
5211 * config/pa/pa.c (pa_modes_tieable_p): Make static.
5212 (TARGET_MODES_TIEABLE_P): Redefine.
5213 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
5214 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
5215 (pdp11_modes_tieable_p): New function.
5216 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
5217 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
5218 (rs6000_modes_tieable_p): New function.
5219 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
5220 * config/powerpcspe/powerpcspe.md: Update comment.
5221 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
5222 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
5223 (TARGET_MODES_TIEABLE_P): Redefine.
5224 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
5225 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
5226 (rl78_modes_tieable_p): New function.
5227 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
5228 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
5229 (rs6000_modes_tieable_p): New function.
5230 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
5231 * config/rs6000/rs6000.md: Update comment.
5232 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
5233 * config/rx/rx.c (rx_modes_tieable_p): New function.
5234 (TARGET_MODES_TIEABLE_P): Redefine.
5235 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
5236 * config/s390/s390.c (s390_modes_tieable_p): New function.
5237 (TARGET_MODES_TIEABLE_P): Redefine.
5238 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
5239 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
5240 (sh_modes_tieable_p): New function.
5241 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
5242 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
5243 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
5244 (sparc_modes_tieable_p): Make static.
5245 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
5246 * config/spu/spu.c (spu_modes_tieable_p): New function.
5247 (TARGET_MODES_TIEABLE_P): Redefine.
5248 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
5249 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
5250 (TARGET_MODES_TIEABLE_P): Redefine.
5251 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
5252 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
5253 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
5254 * config/v850/v850.c (v850_modes_tieable_p): New function.
5255 (TARGET_MODES_TIEABLE_P): Redefine.
5256 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
5257 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
5258 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
5259 (visium_modes_tieable_p): New function.
5260 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
5261 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
5262 (xtensa_modes_tieable_p): New function.
5263 * system.h (MODES_TIEABLE_P): Poison.
5265 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5266 Alan Hayward <alan.hayward@arm.com>
5267 David Sherwood <david.sherwood@arm.com>
5269 * target.def (hard_regno_mode_ok): New hook.
5270 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
5271 (TARGET_HARD_REGNO_MODE_OK): ...this.
5272 * doc/tm.texi.in: Regenerate.
5273 * hooks.h (hook_bool_uint_mode_true): Declare.
5274 * hooks.c (hook_bool_uint_mode_true): New function.
5275 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
5277 * genpreds.c (write_insn_preds_c): Add an include of target.h.
5278 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
5279 instead of HARD_REGNO_MODE_OK.
5280 * caller-save.c: Include target.h.
5281 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
5283 * combine.c (can_combine_p): Likewise.
5284 (combinable_i3pat): Likewise.
5285 (can_change_dest_mode): Likewise.
5286 * expr.c (init_expr_target): Likewise.
5287 (convert_move): Likewise.
5288 (convert_modes): Likewise.
5289 * ira.c (setup_prohibited_class_mode_regs): Likewise.
5290 (setup_prohibited_mode_move_regs): Likewise.
5291 * ira.h (target_ira): Likewise.
5292 * lra-assigns.c (find_hard_regno_for_1): Likewise.
5293 * lra-constraints.c (process_alt_operands): Likewise.
5294 (split_reg): Likewise.
5295 * recog.c (peep2_find_free_register): Likewise.
5296 * ree.c (combine_reaching_defs): Likewise.
5297 * regcprop.c (maybe_mode_change): Likewise.
5298 * reginfo.c (init_reg_sets_1): Likewise.
5299 (choose_hard_reg_mode): Likewise.
5300 (simplifiable_subregs): Likewise.
5301 * regrename.c (check_new_reg_p): Likewise.
5302 * reload.c (find_valid_class): Likewise.
5303 (find_valid_class_1): Likewise.
5304 (reload_inner_reg_of_subreg): Likewise.
5305 (push_reload): Likewise.
5306 (combine_reloads): Likewise.
5307 (find_dummy_reload): Likewise.
5308 (find_reloads): Likewise.
5309 * reload1.c (find_reg): Likewise.
5310 (set_reload_reg): Likewise.
5311 (allocate_reload_reg): Likewise.
5312 (choose_reload_regs): Likewise.
5313 (reload_adjust_reg_for_temp): Likewise.
5314 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
5315 (simplify_subreg_regno): Likewise.
5316 * sel-sched.c (init_regs_for_mode): Likewise.
5317 * varasm.c (make_decl_rtl): Likewise.
5318 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
5319 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
5321 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
5322 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
5323 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5324 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
5325 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
5326 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5327 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
5328 (arc_mode_class): Delete.
5329 (HARD_REGNO_MODE_OK): Delete.
5330 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5331 (arc_hard_regno_mode_ok): Rename old array to...
5332 (arc_hard_regno_mode_ok_modes): ...this.
5333 (arc_conditional_register_usage): Update accordingly.
5334 (arc_mode_class): Make static.
5335 (arc_hard_regno_mode_ok): New function.
5336 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
5337 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
5338 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5339 (arm_hard_regno_mode_ok): Make static.
5340 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
5342 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
5343 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
5344 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
5346 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5347 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
5348 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
5349 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
5350 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
5351 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5352 * config/bfin/predicates.md (valid_reg_operand): Use
5353 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
5354 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
5355 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
5356 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5357 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
5358 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
5359 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5360 (cr16_hard_regno_mode_ok): Make static and return a bool.
5361 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
5362 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5363 (cris_hard_regno_mode_ok): New function.
5364 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
5365 (epiphany_mode_class): Delete.
5366 (HARD_REGNO_MODE_OK): Delete.
5367 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
5368 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5369 (hard_regno_mode_ok): Rename to...
5370 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
5371 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
5372 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
5374 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
5375 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
5376 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5377 (frv_hard_regno_mode_ok): Make static and return a bool.
5378 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
5380 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
5381 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
5382 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
5383 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
5385 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5386 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
5387 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
5388 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
5390 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5391 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
5392 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5393 (ia64_hard_regno_mode_ok): New function.
5394 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
5395 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5396 (iq2000_hard_regno_mode_ok): New function.
5397 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
5398 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5399 (lm32_hard_regno_mode_ok): New function.
5400 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
5401 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
5402 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
5403 instead of HARD_REGNO_MODE_OK.
5404 (m32c_hard_regno_ok): Rename to...
5405 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
5406 (m32c_cannot_change_mode_class): Update accordingly.
5407 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5408 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
5409 (m32r_mode_class): Delete.
5410 (HARD_REGNO_MODE_OK): Delete.
5411 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5412 (m32r_hard_regno_mode_ok): Rename to...
5413 (m32r_hard_regno_modes): ...this.
5414 (m32r_mode_class): Make static.
5415 (m32r_hard_regno_mode_ok): New function.
5416 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
5417 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
5418 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5419 (m68k_hard_regno_mode_ok): Make static.
5420 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
5421 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5422 (mcore_hard_regno_mode_ok): New function.
5423 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
5424 (HARD_REGNO_MODE_OK): Delete.
5425 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
5427 (microblaze_hard_regno_mode_ok_p): ...this and make static.
5428 (microblaze_hard_regno_mode_ok): New function.
5429 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5430 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
5431 (mips_hard_regno_mode_ok): Delete.
5432 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
5433 (mips_hard_regno_mode_ok_p): ...this and make static.
5434 (mips_hard_regno_mode_ok_p): Rename to...
5435 (mips_hard_regno_mode_ok_uncached): ...this.
5436 (mips_hard_regno_mode_ok): New function.
5437 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
5438 of HARD_REGNO_MODE_OK.
5439 (mips_option_override): Update after above name changes.
5440 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5441 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
5442 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
5443 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
5444 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
5445 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5446 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
5447 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
5448 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
5449 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5450 (msp430_hard_regno_mode_ok): Make static and return a bool.
5451 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
5452 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
5453 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
5455 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5456 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
5457 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
5458 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
5459 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
5460 (PA_HARD_REGNO_MODE_OK): ...this
5461 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
5462 (PA_HARD_REGNO_MODE_OK): ...this.
5463 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5464 (pa_hard_regno_mode_ok): New function.
5465 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
5466 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5467 (pdp11_hard_regno_mode_ok): New function.
5468 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
5469 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
5471 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
5473 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5474 (rs6000_hard_regno_mode_ok): Rename to...
5475 (rs6000_hard_regno_mode_ok_uncached): ...this.
5476 (rs6000_init_hard_regno_mode_ok): Update accordingly.
5477 (rs6000_hard_regno_mode_ok): New function.
5478 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
5479 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
5480 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
5481 (riscv_hard_regno_mode_ok): ...this and make static.
5482 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5483 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
5484 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
5485 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5486 (rl78_hard_regno_mode_ok): Make static and return bool.
5487 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
5488 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
5490 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
5491 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5492 (rs6000_hard_regno_mode_ok): Rename to...
5493 (rs6000_hard_regno_mode_ok_uncached): ...this.
5494 (rs6000_init_hard_regno_mode_ok): Update accordingly.
5495 (rs6000_hard_regno_mode_ok): New function.
5496 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
5497 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
5498 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5499 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
5500 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
5501 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
5502 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5503 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
5504 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
5505 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5506 (sh_hard_regno_mode_ok): Make static.
5507 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
5508 instead of HARD_REGNO_MODE_OK.
5509 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
5510 (sparc_mode_class): Delete.
5511 (HARD_REGNO_MODE_OK): Delete.
5512 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5513 (hard_regno_mode_classes): Make static.
5514 (sparc_mode_class): Likewise.
5515 (sparc_hard_regno_mode_ok): New function.
5516 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
5517 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
5518 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
5520 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5521 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
5522 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
5523 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
5524 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
5525 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5526 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
5527 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
5528 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5529 (visium_hard_regno_mode_ok): New function.
5530 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
5531 instead of HARD_REGNO_MODE_OK.
5532 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
5533 (HARD_REGNO_MODE_OK): Delete.
5534 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
5535 (xtensa_hard_regno_mode_ok_p): ...this and make static.
5536 (xtensa_option_override): Update accordingly.
5537 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5538 (xtensa_hard_regno_mode_ok): New function.
5539 * system.h (HARD_REGNO_MODE_OK): Poison.
5541 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5542 Alan Hayward <alan.hayward@arm.com>
5543 David Sherwood <david.sherwood@arm.com>
5545 * target.def (hard_regno_call_part_clobbered): New hook.
5546 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
5547 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
5548 * doc/tm.texi: Regenerate.
5549 * hooks.h (hook_bool_uint_mode_false): Declare.
5550 * hooks.c (hook_bool_uint_mode_false): New function.
5551 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5552 * cselib.c (cselib_process_insn): Use
5553 targetm.hard_regno_call_part_clobbered instead of
5554 HARD_REGNO_CALL_PART_CLOBBERED.
5555 * ira-conflicts.c (ira_build_conflicts): Likewise.
5556 * ira-costs.c (ira_tune_allocno_costs): Likewise.
5557 * lra-constraints.c (need_for_call_save_p): Likewise.
5558 * lra-lives.c: Include target.h.
5559 (check_pseudos_live_through_calls): Use
5560 targetm.hard_regno_call_part_clobbered instead of
5561 HARD_REGNO_CALL_PART_CLOBBERED.
5562 * regcprop.c: Include target.h.
5563 (copyprop_hardreg_forward_1): Use
5564 targetm.hard_regno_call_part_clobbered instead of
5565 HARD_REGNO_CALL_PART_CLOBBERED.
5566 * reginfo.c (choose_hard_reg_mode): Likewise.
5567 * regrename.c (check_new_reg_p): Likewise.
5568 * reload.c (find_equiv_reg): Likewise.
5569 * reload1.c (emit_reload_insns): Likewise.
5570 * sched-deps.c (deps_analyze_insn): Likewise.
5571 * sel-sched.c (init_regs_for_mode): Likewise.
5572 (mark_unavailable_hard_regs): Likewise.
5573 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
5574 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5575 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
5577 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5578 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5579 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
5581 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
5583 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5584 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5585 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
5587 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5588 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5589 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
5591 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5592 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
5594 * config/powerpcspe/powerpcspe.c
5595 (rs6000_hard_regno_call_part_clobbered): New function.
5596 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5597 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5598 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
5600 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5601 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5602 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
5604 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5605 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5606 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
5608 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5609 Alan Hayward <alan.hayward@arm.com>
5610 David Sherwood <david.sherwood@arm.com>
5612 * rtl.h (subreg_memory_offset): Declare.
5613 * emit-rtl.c (subreg_memory_offset): New function.
5614 * expmed.c (store_bit_field_1): Use it.
5615 * expr.c (undefined_operand_subword_p): Likewise.
5616 * simplify-rtx.c (simplify_subreg): Likewise.
5618 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
5620 PR rtl-optimization/57448
5623 * optabs.c (expand_atomic_load): Place compiler memory barriers if
5624 using atomic_load pattern.
5625 (expand_atomic_store): Likewise.
5627 2017-09-04 Jakub Jelinek <jakub@redhat.com>
5630 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
5631 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
5632 handling. Use replace_call_with_value with NULL instead of
5633 gsi_replace, unlink_stmt_vdef and release_defs.
5635 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
5638 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
5640 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5643 * rtl.h (emit_library_call_value_1): Declare.
5644 (emit_library_call): Replace declaration with a series of overloads.
5645 Remove the parameter count argument.
5646 (emit_library_call_value): Likewise.
5647 * calls.c (emit_library_call_value_1): Make global. Replace varargs
5648 with an "rtx_mode_t *".
5649 (emit_library_call_value): Delete.
5650 (emit_library_call): Likewise.
5651 * asan.c (asan_emit_stack_protection): Update calls accordingly.
5652 (asan_emit_allocas_unpoison): Likewise.
5653 * builtins.c (expand_builtin_powi): Likewise.
5654 (expand_asan_emit_allocas_unpoison): Likewise.
5655 * cfgexpand.c (expand_main_function): Likewise.
5656 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
5657 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
5658 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
5659 * config/arm/arm.c (arm_trampoline_init): Likewise.
5660 (arm_call_tls_get_addr): Likewise.
5661 (arm_expand_divmod_libfunc): Likewise.
5662 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
5663 (smulsi3_highpart): Likewise.
5664 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
5665 (c6x_expand_compare): Likewise.
5666 (c6x_expand_movmem): Likewise.
5667 * config/frv/frv.c (frv_trampoline_init): Likewise.
5668 * config/i386/i386.c (ix86_trampoline_init): Likewise.
5669 (ix86_expand_divmod_libfunc): Likewise.
5670 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
5671 (ia64_expand_compare): Likewise.
5672 (ia64_profile_hook): Likewise.
5673 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
5674 (nonlocal_goto): Likewise.
5675 (restore_stack_nonlocal): Likewise.
5676 * config/m32r/m32r.c (block_move_call): Likewise.
5677 (m32r_trampoline_init): Likewise.
5678 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
5679 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
5680 (m68k_call_m68k_read_tp): Likewise.
5681 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
5682 (microblaze_expand_divide): Likewise.
5683 * config/mips/mips.h (mips_args): Likewise.
5684 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
5685 (MIPS_ICACHE_SYNC): Likewise.
5686 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
5687 (nios2_trampoline_init): Likewise.
5688 * config/pa/pa.c (hppa_tls_call): Likewise.
5689 (pa_trampoline_init): Likewise.
5690 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
5691 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
5692 (expand_strn_compare): Likewise.
5693 (rs6000_generate_compare): Likewise.
5694 (rs6000_expand_float128_convert): Likewise.
5695 (output_profile_hook): Likewise.
5696 (rs6000_trampoline_init): Likewise.
5697 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
5698 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
5699 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
5700 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
5701 (rs6000_generate_compare): Likewise.
5702 (rs6000_expand_float128_convert): Likewise.
5703 (output_profile_hook): Likewise.
5704 (rs6000_trampoline_init): Likewise.
5705 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
5706 * config/sh/sh.c (sh_trampoline_init): Likewise.
5707 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
5708 (sparc_emit_float_lib_cmp): Likewise.
5709 (sparc32_initialize_trampoline): Likewise.
5710 (sparc64_initialize_trampoline): Likewise.
5711 (sparc_profile_hook): Likewise.
5712 * config/spu/spu.c (ea_load_store): Likewise.
5713 * config/spu/spu.md (floatunssidf2): Likewise.
5714 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
5715 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
5716 * config/visium/visium.c (expand_block_move_4): Likewise.
5717 (expand_block_move_2): Likewise.
5718 (expand_block_move_1): Likewise.
5719 (expand_block_set_4): Likewise.
5720 (expand_block_set_2): Likewise.
5721 (expand_block_set_1): Likewise.
5722 (visium_trampoline_init): Likewise.
5723 (visium_profile_hook): Likewise.
5724 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
5725 (xtensa_setup_frame_addresses): Likewise.
5726 (xtensa_trampoline_init): Likewise.
5727 * except.c (sjlj_emit_function_enter): Likewise.
5728 (sjlj_emit_function_exit): Likewise.
5729 * explow.c (allocate_dynamic_stack_space): Likewise.
5730 (probe_stack_range): Likewise.
5731 * expr.c (convert_mode_scalar): Likewise.
5732 * optabs.c (expand_binop): Likewise.
5733 (expand_twoval_binop_libfunc): Likewise.
5734 (expand_unop): Likewise.
5735 (prepare_cmp_insn): Likewise.
5736 (prepare_float_lib_cmp): Likewise.
5737 (expand_float): Likewise.
5738 (expand_fix): Likewise.
5739 (expand_fixed_convert): Likewise.
5740 (maybe_emit_sync_lock_test_and_set): Likewise.
5741 (expand_atomic_compare_and_swap): Likewise.
5742 (expand_mem_thread_fence): Likewise.
5743 (expand_atomic_fetch_op): Likewise.
5745 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
5747 * doc/generic.texi (OpenACC): Adjust URL.
5748 * doc/invoke.texi (C Dialect Options): Ditto.
5750 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
5752 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
5753 predicate for operand 1. Add (m,<S>) constraint.
5754 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
5755 Prevent memory operand 1 with register operand 2.
5757 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
5759 PR rtl-optimization/82024
5760 * combine.c (try_combine): If the combination result is a PARALLEL,
5761 and we only need to retain the SET in there that would be placed
5762 at I2, check that we can place that at I3 instead, before doing so.
5764 2017-09-01 Jakub Jelinek <jakub@redhat.com>
5767 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
5769 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
5770 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
5773 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
5774 Jeff Law <law@redhat.com>
5776 * varasm.c (bss_initializer_p): Do not put constants into .bss
5777 (categorize_decl_for_section): Handle bss_initializer_p returning
5778 false when DECL_INITIAL is NULL.
5780 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5783 * config/s390/s390.c (s390_can_inline_p): New function.
5785 2017-09-01 Jeff Law <law@redhat.com>
5787 PR tree-optimization/82052
5788 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
5789 Always initialize the returned slot after a hash table miss
5790 when INSERT is true.
5792 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
5794 * config/s390/s390.md (mem_signal_fence): Remove.
5795 * doc/md.texi (mem_signal_fence): Remove.
5796 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
5798 * target-insns.def (mem_signal_fence): Remove.
5800 2017-09-01 Jakub Jelinek <jakub@redhat.com>
5803 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
5806 * asan.c (create_odr_indicator): Strip name encoding from assembler
5807 name before appending it after __odr_asan_.
5809 2017-09-01 Martin Liska <mliska@suse.cz>
5811 PR tree-optimization/82059
5812 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
5813 frequency only when an edge is redirected.
5815 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5817 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
5818 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
5819 (arc_conditional_register_usage): Remove ARC600 lp_count
5821 (arc_file_start): Emit Tag_ARC_CPU_variation.
5822 (arc_can_use_doloop_p): New conditions to use ZOLs.
5823 (hwloop_fail): New function.
5824 (hwloop_optimize): Likewise.
5825 (hwloop_pattern_reg): Likewise.
5826 (arc_doloop_hooks): New struct, to be used with reorg_loops.
5827 (arc_reorg_loops): New function, calls reorg_loops.
5828 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
5829 (arc600_corereg_hazard): Remove ZOL checking, case handled by
5831 (arc_loop_hazard): Remove function, functionality moved into
5833 (arc_hazard): Remove arc_loop_hazard call.
5834 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
5835 into hwloop_optimize.
5836 (arc_label_align): Remove ZOL handling.
5837 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
5838 * config/arc/arc.md (doloop_begin): Remove pattern.
5839 (doloop_begin_i): Likewise.
5840 (doloop_end_i): Likewise.
5841 (doloop_fallback): Likewise.
5842 (doloop_fallback_m): Likewise.
5843 (doloop_end): Reimplement expand.
5844 (arc_lp): New pattern for LP instruction.
5845 (loop_end): New pattern.
5846 (loop_fail): Likewise.
5847 (decrement_and_branch_until_zero): Likewise.
5848 * config/arc/arc.opt (mlpc-width): New option.
5849 * doc/invoke.texi (mlpc-width): Document option.
5851 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5853 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
5854 (arc_ccfsm_advance): Fix checking for delay slots.
5855 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
5857 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5859 * config/arc/arc.md (movqi_insn): Add stores to save constant long
5861 (movhi_insn): Update store instruction constraint which are saving
5862 6-bit short immediates.
5863 (movsi_insn): Consider also short scaled load operations.
5864 (zero_extendhisi2_i): Use Usd constraint instead of T.
5865 (extendhisi2_i): Add q constraint.
5866 (arc_clzsi2): Add type and length attributes.
5867 (arc_ctzsi2): Likewise.
5868 * config/arc/constraints.md (Usc): Update constraint, the
5869 assembler can parse two relocations for a single instruction.
5871 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5873 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
5874 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
5876 2017-08-31 Olivier Hainque <hainque@adacore.com>
5878 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
5879 match as powerpc-wrs-vxworks*.
5881 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
5883 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
5884 register constraint for by-element operand.
5885 (aarch64_mls_elt_merge<mode>): Likewise.
5887 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
5889 * config/arc/arc.c (arc_can_follow_jump): Check for short
5892 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
5894 * config.gcc: Use g.opt for arc.
5895 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
5896 functionality moved to ...
5897 (legitimate_scaled_address_p): New function, ...here.
5898 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
5899 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
5900 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
5902 (arc_override_options): Handle G option.
5903 (arc_output_pic_addr_const): Correct function definition.
5904 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
5905 (arc_decl_anon_ns_mem_p): Delete.
5906 (arc_in_small_data_p): Overhaul this function to take into
5907 consideration the value given via G option.
5908 (arc_rewrite_small_data_1): Renamed and corrected old
5909 arc_rewrite_small_data function.
5910 (arc_rewrite_small_data): New function.
5911 (small_data_pattern): Don't use pic_offset_table_rtx.
5912 * config/arc/arc.h (CC1_SPEC): Recognize G option.
5913 * config/arc/simdext.md (movmisalignv2hi): Use
5914 prepare_move_operands function.
5916 (movmisalign*): Likewise.
5917 * doc/invoke.texi (ARC options): Document -G option.
5919 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
5921 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
5923 * config/arc/arc.c (arc_print_operand): Output scalled address for
5924 sdata whenever is possible.
5925 (arc_in_small_data_p): Allow sdata for 64bit datum when double
5926 load/stores are available.
5927 (compact_sda_memory_operand): Check for the alignment required by
5928 code density instructions.
5929 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
5931 * config/arc/constraints.md (Usd): Update constraint.
5932 (Us0): New constraint.
5933 (Usc): Update constraint.
5935 2017-08-31 Richard Biener <rguenther@suse.de>
5938 * dwarf2out.c (dwarf2out_early_global_decl): Process each
5941 2017-08-31 Tamar Christina <tamar.christina@arm.com>
5943 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
5944 Resize type_signature.
5946 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
5947 Alan Hayward <alan.hayward@arm.com>
5948 David Sherwood <david.sherwood@arm.com>
5950 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
5951 subregs whose inner modes can be stored in GPRs.
5952 (aarch64_classify_index): Likewise.
5954 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
5955 Alan Hayward <alan.hayward@arm.com>
5956 David Sherwood <david.sherwood@arm.com>
5958 * config/aarch64/iterators.md (V_cmp_result): Rename to...
5959 (V_INT_EQUIV): ...this.
5960 (v_cmp_result): Rename to...
5961 (v_int_equiv): ...this.
5962 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
5963 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
5964 (copysign<mode>3): Likewise.
5965 (aarch64_simd_bsl<mode>_internal): Likewise.
5966 (aarch64_simd_bsl<mode>): Likewise.
5967 (vec_cmp<mode><mode>): Likewise.
5968 (vcond<mode><mode>): Likewise.
5969 (vcond<v_cmp_mixed><mode>): Likewise.
5970 (vcondu<mode><v_cmp_mixed>): Likewise.
5971 (aarch64_cm<optab><mode>): Likewise.
5972 (aarch64_cmtst<mode>): Likewise.
5973 (aarch64_fac<optab><mode>): Likewise.
5974 (vec_perm_const<mode>): Likewise.
5975 (vcond_mask_<mode><v_cmp_result>): Rename to...
5976 (vcond_mask_<mode><v_int_equiv>): ...this.
5977 (vec_cmp<mode><v_cmp_result>): Rename to...
5978 (vec_cmp<mode><v_int_equiv>): ...this.
5980 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
5981 Alan Hayward <alan.hayward@arm.com>
5982 David Sherwood <david.sherwood@arm.com>
5984 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
5986 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
5987 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
5988 (UNSPEC_LD4_DREG): New unspecs.
5989 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
5990 (aarch64_ld2<mode>_dreg_be): Replace with...
5991 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
5993 (aarch64_ld3<mode>_dreg_le)
5994 (aarch64_ld3<mode>_dreg_be): Replace with...
5995 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
5997 (aarch64_ld4<mode>_dreg_le)
5998 (aarch64_ld4<mode>_dreg_be): Replace with...
5999 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
6002 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6004 PR tree-optimization/81987
6005 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
6006 insert an initializer in a location not dominated by the stride
6009 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
6011 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
6012 on the entire header of the finally block in the fallthru case.
6014 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
6016 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
6018 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
6020 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
6021 rs6000_emit_move_from_cr and call renamed function.
6022 (rs6000_emit_prologue): Call renamed functions.
6023 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
6024 movesi_from_cr, remove volatile CRs.
6026 2017-08-30 Jon Beniston <jon@beniston.com>
6027 Richard Biener <rguenther@suse.de>
6029 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
6030 of VECTOR_MODE_P check.
6031 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
6032 element vector types.
6034 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6036 * df.h (df_read_modify_subreg_p): Remove in favor of...
6037 * rtl.h (read_modify_subreg_p): ...this new function. Take a
6038 const_rtx instead of an rtx.
6039 * cprop.c (local_cprop_find_used_regs): Update accordingly.
6040 * df-problems.c (df_word_lr_mark_ref): Likewise.
6041 * ira-lives.c (mark_pseudo_reg_live): Likewise.
6042 (mark_pseudo_reg_dead): Likewise.
6043 (mark_ref_dead): Likewise.
6044 * reginfo.c (init_subregs_of_mode): Likewise.
6045 * sched-deps.c (sched_analyze_1): Likewise.
6046 * df-scan.c (df_def_record_1): Likewise.
6047 (df_uses_record): Likewise.
6048 (df_read_modify_subreg_p): Remove in favor of...
6049 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
6050 const_rtx instead of an rtx.
6052 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6053 Alan Hayward <alan.hayward@arm.com>
6054 David Sherwood <david.sherwood@arm.com>
6056 * rtl.h (partial_subreg_p): New function.
6057 * caller-save.c (save_call_clobbered_regs): Use it.
6058 * calls.c (expand_call): Likewise.
6059 * combine.c (combinable_i3pat): Likewise.
6060 (simplify_set): Likewise.
6061 (make_extraction): Likewise.
6062 (make_compound_operation_int): Likewise.
6063 (gen_lowpart_or_truncate): Likewise.
6064 (force_to_mode): Likewise.
6065 (make_field_assignment): Likewise.
6066 (reg_truncated_to_mode): Likewise.
6067 (record_truncated_value): Likewise.
6068 (move_deaths): Likewise.
6069 * cse.c (record_jump_cond): Likewise.
6070 (cse_insn): Likewise.
6071 * cselib.c (cselib_lookup_1): Likewise.
6072 * expmed.c (extract_bit_field_using_extv): Likewise.
6073 * function.c (assign_parm_setup_reg): Likewise.
6074 * ifcvt.c (noce_convert_multiple_sets): Likewise.
6075 * ira-build.c (create_insn_allocnos): Likewise.
6076 * lra-coalesce.c (merge_pseudos): Likewise.
6077 * lra-constraints.c (match_reload): Likewise.
6078 (simplify_operand_subreg): Likewise.
6079 (curr_insn_transform): Likewise.
6080 * lra-lives.c (process_bb_lives): Likewise.
6081 * lra.c (new_insn_reg): Likewise.
6082 (lra_substitute_pseudo): Likewise.
6083 * regcprop.c (mode_change_ok): Likewise.
6084 (maybe_mode_change): Likewise.
6085 (copyprop_hardreg_forward_1): Likewise.
6086 * reload.c (push_reload): Likewise.
6087 (find_reloads): Likewise.
6088 (find_reloads_subreg_address): Likewise.
6089 * reload1.c (alter_reg): Likewise.
6090 (eliminate_regs_1): Likewise.
6091 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6093 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
6095 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
6096 back to if statements, including unpack.
6098 2017-08-30 Martin Liska <mliska@suse.cz>
6101 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
6103 (func_checker::compare_asm_inputs_outputs): ... this function.
6104 (func_checker::compare_gimple_asm): Use the function to compare
6105 also ASM constrains.
6106 * ipa-icf-gimple.h: Rename the function.
6108 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6109 Alan Hayward <alan.hayward@arm.com>
6110 David Sherwood <david.sherwood@arm.com>
6112 * coretypes.h (complex_mode): New type.
6113 * gdbhooks.py (build_pretty_printer): Handle it.
6114 * machmode.h (complex_mode): New class.
6115 (complex_mode::includes_p): New function.
6116 (is_complex_int_mode): Likewise.
6117 (is_complex_float_mode): Likewise.
6118 * genmodes.c (get_mode_class): Handle complex mode classes.
6119 * function.c (expand_function_end): Use is_complex_int_mode.
6121 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6122 Alan Hayward <alan.hayward@arm.com>
6123 David Sherwood <david.sherwood@arm.com>
6125 * coretypes.h (scalar_mode_pod): New typedef.
6126 * gdbhooks.py (build_pretty_printer): Handle it.
6127 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
6128 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
6129 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
6130 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
6133 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6134 Alan Hayward <alan.hayward@arm.com>
6135 David Sherwood <david.sherwood@arm.com>
6137 * machmode.h (mode_for_vector): Take a scalar_mode instead
6139 * stor-layout.c (mode_for_vector): Likewise.
6140 * explow.c (promote_mode): Use as_a <scalar_mode>.
6141 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
6143 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6144 Alan Hayward <alan.hayward@arm.com>
6145 David Sherwood <david.sherwood@arm.com>
6147 * target.def (preferred_simd_mode): Take a scalar_mode
6148 instead of a machine_mode.
6149 * targhooks.h (default_preferred_simd_mode): Likewise.
6150 * targhooks.c (default_preferred_simd_mode): Likewise.
6151 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
6152 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
6153 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
6154 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
6155 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
6156 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
6157 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
6158 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
6160 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
6161 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
6162 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
6163 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
6164 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
6165 * doc/tm.texi: Regenerate.
6166 * optabs-query.c (can_vec_mask_load_store_p): Return false for
6169 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6170 Alan Hayward <alan.hayward@arm.com>
6171 David Sherwood <david.sherwood@arm.com>
6173 * target.def (scalar_mode_supported_p): Take a scalar_mode
6174 instead of a machine_mode.
6175 * targhooks.h (default_scalar_mode_supported_p): Likewise.
6176 * targhooks.c (default_scalar_mode_supported_p): Likewise.
6177 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
6178 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
6179 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
6180 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
6181 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
6182 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
6183 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
6184 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
6185 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
6186 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
6187 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
6188 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
6190 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
6191 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
6192 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
6193 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
6194 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
6196 * doc/tm.texi: Regenerate.
6198 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6199 Alan Hayward <alan.hayward@arm.com>
6200 David Sherwood <david.sherwood@arm.com>
6202 * coretypes.h (opt_scalar_mode): New typedef.
6203 * gdbhooks.py (build_pretty_printers): Handle it.
6204 * machmode.h (mode_iterator::get_2xwider): Add overload for
6206 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
6208 * expr.c (convert_mode_scalar): Likewise.
6209 * omp-low.c (omp_clause_aligned_alignment): Likewise.
6210 * optabs.c (expand_float): Likewise.
6211 (expand_fix): Likewise.
6212 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6214 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6215 Alan Hayward <alan.hayward@arm.com>
6216 David Sherwood <david.sherwood@arm.com>
6218 * optabs.c (expand_float): Explicitly check for scalars before
6219 using a branching expansion.
6220 (expand_fix): Likewise.
6222 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6223 Alan Hayward <alan.hayward@arm.com>
6224 David Sherwood <david.sherwood@arm.com>
6226 * expr.c (convert_mode): Split scalar handling out into...
6227 (convert_mode_scalar): ...this new function. Treat the modes
6230 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6231 Alan Hayward <alan.hayward@arm.com>
6232 David Sherwood <david.sherwood@arm.com>
6234 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
6236 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
6238 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6239 Alan Hayward <alan.hayward@arm.com>
6240 David Sherwood <david.sherwood@arm.com>
6242 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
6243 rather than a machine_mode.
6244 (fixed_from_string): Likewise.
6245 (fixed_convert): Likewise.
6246 (fixed_convert_from_int): Likewise.
6247 (fixed_convert_from_real): Likewise.
6248 (real_convert_from_fixed): Likewise.
6249 * fixed-value.c (fixed_from_double_int): Likewise.
6250 (fixed_from_string): Likewise.
6251 (fixed_convert): Likewise.
6252 (fixed_convert_from_int): Likewise.
6253 (fixed_convert_from_real): Likewise.
6254 (real_convert_from_fixed): Likewise.
6255 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
6257 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6258 Alan Hayward <alan.hayward@arm.com>
6259 David Sherwood <david.sherwood@arm.com>
6261 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
6262 of separate mode class checks. Do not allow vector modes here.
6263 (immed_wide_int_const): Use as_a <scalar_mode>.
6264 * explow.c (trunc_int_for_mode): Likewise.
6265 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
6266 (wi::shwi): Likewise.
6267 (wi::min_value): Likewise.
6268 (wi::max_value): Likewise.
6269 * dwarf2out.c (loc_descriptor): Likewise.
6270 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
6273 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6274 Alan Hayward <alan.hayward@arm.com>
6275 David Sherwood <david.sherwood@arm.com>
6277 * tree.h (SCALAR_TYPE_MODE): New macro.
6278 * expr.c (expand_expr_addr_expr_1): Use it.
6279 (expand_expr_real_2): Likewise.
6280 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
6281 (fold_convert_const_fixed_from_int): Likewise.
6282 (fold_convert_const_fixed_from_real): Likewise.
6283 (native_encode_fixed): Likewise
6284 (native_encode_complex): Likewise
6285 (native_encode_vector): Likewise.
6286 (native_interpret_fixed): Likewise.
6287 (native_interpret_real): Likewise.
6288 (native_interpret_complex): Likewise.
6289 (native_interpret_vector): Likewise.
6290 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
6291 (simd_clone_adjust_argument_types): Likewise.
6292 (simd_clone_init_simd_arrays): Likewise.
6293 (simd_clone_adjust): Likewise.
6294 * stor-layout.c (layout_type): Likewise.
6295 * tree.c (build_minus_one_cst): Likewise.
6296 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
6297 * tree-inline.c (estimate_move_cost): Likewise.
6298 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
6299 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
6300 (vectorizable_reduction): Likewise.
6301 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
6302 (vect_recog_mixed_size_cond_pattern): Likewise.
6303 (check_bool_pattern): Likewise.
6304 (adjust_bool_pattern): Likewise.
6305 (search_type_for_mask_1): Likewise.
6306 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
6307 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6308 (vectorizable_load): Likewise.
6309 (vectorizable_store): Likewise.
6310 * ubsan.c (ubsan_encode_value): Likewise.
6311 * varasm.c (output_constant): Likewise.
6313 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6314 Alan Hayward <alan.hayward@arm.com>
6315 David Sherwood <david.sherwood@arm.com>
6317 * coretypes.h (scalar_mode): New class.
6318 * machmode.h (scalar_mode): Likewise.
6319 (scalar_mode::includes_p): New function.
6320 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
6321 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
6322 * genmodes.c (get_mode_class): Handle remaining scalar modes.
6323 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
6324 * expmed.c (store_bit_field_1): Likewise.
6325 (extract_bit_field_1): Likewise.
6326 * expr.c (write_complex_part): Likewise.
6327 (read_complex_part): Likewise.
6328 (emit_move_complex_push): Likewise.
6329 (expand_expr_real_2): Likewise.
6330 * function.c (assign_parm_setup_reg): Likewise.
6331 (assign_parms_unsplit_complex): Likewise.
6332 * optabs.c (expand_binop): Likewise.
6333 * rtlanal.c (subreg_get_info): Likewise.
6334 * simplify-rtx.c (simplify_immed_subreg): Likewise.
6335 * varasm.c (output_constant_pool_2): Likewise.
6337 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6338 Alan Hayward <alan.hayward@arm.com>
6339 David Sherwood <david.sherwood@arm.com>
6341 * expmed.c (extract_high_half): Use scalar_int_mode and remove
6343 (expmed_mult_highpart_optab): Likewise.
6344 (expmed_mult_highpart): Likewise.
6346 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6347 Alan Hayward <alan.hayward@arm.com>
6348 David Sherwood <david.sherwood@arm.com>
6350 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
6351 instead of a machine_mode.
6352 (builtin_memset_read_str): Likewise.
6353 * builtins.c (c_readstr): Likewise.
6354 (builtin_memcpy_read_str): Likewise.
6355 (builtin_strncpy_read_str): Likewise.
6356 (builtin_memset_read_str): Likewise.
6357 (builtin_memset_gen_str): Likewise.
6358 (expand_builtin_signbit): Use scalar_int_mode for local variables.
6359 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
6360 instead of a machine_mode.
6361 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
6363 (make_extraction): Likewise.
6364 (try_widen_shift_mode): Take and return scalar_int_modes instead
6366 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
6367 a scalar_int_mode instead of a machine_mode.
6368 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
6369 (avr_addr_space_pointer_mode): Likewise.
6370 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
6371 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
6372 (msp430_unwind_word_mode): Likewise.
6373 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
6374 (spu_addr_space_pointer_mode): Likewise.
6375 (spu_addr_space_address_mode): Likewise.
6376 (spu_libgcc_cmp_return_mode): Likewise.
6377 (spu_libgcc_shift_count_mode): Likewise.
6378 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
6379 (rl78_addr_space_pointer_mode): Likewise.
6380 (fl78_unwind_word_mode): Likewise.
6381 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
6383 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
6384 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
6385 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
6386 (mips_valid_pointer_mode): Likewise.
6387 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
6388 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
6389 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
6391 (ft32_addr_space_address_mode): Likewise.
6392 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
6393 scalar_int_mode instead of a machine_mode.
6394 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
6396 (m32c_addr_space_address_mode): Likewise.
6397 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
6398 (rs6000_eh_return_filter_mode): Likewise.
6399 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
6400 (rs6000_eh_return_filter_mode): Likewise.
6401 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
6402 (s390_libgcc_shift_count_mode): Likewise.
6403 (s390_unwind_word_mode): Likewise.
6404 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
6406 * target.def (mode_rep_extended): Likewise.
6407 (valid_pointer_mode): Likewise.
6408 (addr_space.valid_pointer_mode): Likewise.
6409 (eh_return_filter_mode): Return a scalar_int_mode rather than
6411 (libgcc_cmp_return_mode): Likewise.
6412 (libgcc_shift_count_mode): Likewise.
6413 (unwind_word_mode): Likewise.
6414 (addr_space.pointer_mode): Likewise.
6415 (addr_space.address_mode): Likewise.
6416 * doc/tm.texi: Regenerate.
6417 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
6419 (do_jump): Use scalar_int_mode for local variables.
6420 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
6421 rather than a machine_mode.
6422 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
6423 (scompare_loc_descriptor_wide): Likewise.
6424 (scompare_loc_descriptor_narrow): Likewise.
6425 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
6427 * except.c (sjlj_emit_dispatch_table): Likewise.
6428 (expand_builtin_eh_copy_values): Likewise.
6429 * explow.c (convert_memory_address_addr_space_1): Likewise.
6430 Take a scalar_int_mode rather than a machine_mode.
6431 (convert_memory_address_addr_space): Take a scalar_int_mode rather
6432 than a machine_mode.
6433 (memory_address_addr_space): Use scalar_int_mode for local variables.
6434 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
6435 rather than a machine_mode.
6436 * expmed.c (mask_rtx): Likewise.
6437 (init_expmed_one_conv): Likewise.
6438 (expand_mult_highpart_adjust): Likewise.
6439 (extract_high_half): Likewise.
6440 (expmed_mult_highpart_optab): Likewise.
6441 (expmed_mult_highpart): Likewise.
6442 (expand_smod_pow2): Likewise.
6443 (expand_sdiv_pow2): Likewise.
6444 (emit_store_flag_int): Likewise.
6445 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
6447 (extract_low_bits): Likewise.
6448 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
6450 * expr.c (pieces_addr::adjust): Likewise.
6451 (can_store_by_pieces): Likewise.
6452 (store_by_pieces): Likewise.
6453 (clear_by_pieces_1): Likewise.
6454 (expand_expr_addr_expr_1): Likewise.
6455 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
6456 (expand_expr_real_1): Likewise.
6457 (try_casesi): Likewise.
6458 * final.c (shorten_branches): Likewise.
6459 * fold-const.c (fold_convert_const_int_from_fixed): Change the
6460 type of "mode" to machine_mode.
6461 * internal-fn.c (expand_arith_overflow_result_store): Take a
6462 scalar_int_mode rather than a machine_mode.
6463 (expand_mul_overflow): Use scalar_int_mode for local variables.
6464 * loop-doloop.c (doloop_modify): Likewise.
6465 (doloop_optimize): Likewise.
6466 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
6467 than a machine_mode.
6468 (expand_doubleword_shift_condmove): Likewise.
6469 (expand_doubleword_shift): Likewise.
6470 (expand_doubleword_clz): Likewise.
6471 (expand_doubleword_popcount): Likewise.
6472 (expand_doubleword_parity): Likewise.
6473 (expand_absneg_bit): Use scalar_int_mode for local variables.
6474 (prepare_float_lib_cmp): Likewise.
6475 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
6476 rather than a machine_mode.
6477 (convert_memory_address_addr_space): Likewise.
6478 (get_mode_bounds): Likewise.
6479 (get_address_mode): Return a scalar_int_mode rather than a
6481 * rtlanal.c (get_address_mode): Likewise.
6482 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
6483 than a machine_mode.
6484 * targhooks.c (default_mode_rep_extended): Likewise.
6485 (default_valid_pointer_mode): Likewise.
6486 (default_addr_space_valid_pointer_mode): Likewise.
6487 (default_eh_return_filter_mode): Return a scalar_int_mode rather
6488 than a machine_mode.
6489 (default_libgcc_cmp_return_mode): Likewise.
6490 (default_libgcc_shift_count_mode): Likewise.
6491 (default_unwind_word_mode): Likewise.
6492 (default_addr_space_pointer_mode): Likewise.
6493 (default_addr_space_address_mode): Likewise.
6494 * targhooks.h (default_eh_return_filter_mode): Likewise.
6495 (default_libgcc_cmp_return_mode): Likewise.
6496 (default_libgcc_shift_count_mode): Likewise.
6497 (default_unwind_word_mode): Likewise.
6498 (default_addr_space_pointer_mode): Likewise.
6499 (default_addr_space_address_mode): Likewise.
6500 (default_mode_rep_extended): Take a scalar_int_mode rather than
6502 (default_valid_pointer_mode): Likewise.
6503 (default_addr_space_valid_pointer_mode): Likewise.
6504 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
6506 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
6507 rather than a machine_mode.
6508 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
6509 for local variables.
6510 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6511 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
6512 than a machine_mode.
6514 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6515 Alan Hayward <alan.hayward@arm.com>
6516 David Sherwood <david.sherwood@arm.com>
6518 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
6519 the mode argument to scalar_int_mode.
6520 (do_jump_by_parts_zero_rtx): Likewise.
6521 (do_jump_by_parts_equality_rtx): Likewise.
6522 (do_jump_by_parts_greater): Take a mode argument.
6523 (do_jump_by_parts_equality): Likewise.
6524 (do_jump_1): Update calls accordingly.
6526 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6527 Alan Hayward <alan.hayward@arm.com>
6528 David Sherwood <david.sherwood@arm.com>
6530 * is-a.h (safe_dyn_cast): New function.
6531 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
6532 (jump_table_for_label): Likewise.
6533 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
6534 instead of an rtx_insn *.
6535 (shorten_branches): Use dyn_cast instead of LABEL_P and
6536 JUMP_TABLE_DATA_P. Use jump_table_for_label and
6537 rtx_jump_table_data::get_data_mode.
6538 (final_scan_insn): Likewise.
6540 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6541 Alan Hayward <alan.hayward@arm.com>
6542 David Sherwood <david.sherwood@arm.com>
6544 * combine.c (try_combine): Use is_a <scalar_int_mode> when
6545 trying to combine a full-register integer set with a subreg
6548 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6549 Alan Hayward <alan.hayward@arm.com>
6550 David Sherwood <david.sherwood@arm.com>
6552 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
6553 that is always either address_mode or pointer_mode.
6555 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6556 Alan Hayward <alan.hayward@arm.com>
6557 David Sherwood <david.sherwood@arm.com>
6559 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
6560 when the two are known to be equal.
6562 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6563 Alan Hayward <alan.hayward@arm.com>
6564 David Sherwood <david.sherwood@arm.com>
6566 * simplify-rtx.c (simplify_const_unary_operation): Use
6567 is_a <scalar_int_mode> instead of checking for a nonzero
6568 precision. Forcibly convert op_mode to a scalar_int_mode
6569 in that case. More clearly differentiate the operand and
6570 result modes and use the former when deciding what the value
6571 of a count-bits operation should be. Use is_int_mode instead
6572 of checking for a MODE_INT. Remove redundant check for whether
6573 this mode has a zero precision.
6575 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6576 Alan Hayward <alan.hayward@arm.com>
6577 David Sherwood <david.sherwood@arm.com>
6579 * optabs.c (widen_leading): Change the type of the mode argument
6580 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
6581 (widen_bswap): Likewise.
6582 (expand_parity): Likewise.
6583 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
6584 (expand_ffs): Likewise.
6585 (epand_unop): Check for scalar integer modes before calling the
6588 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6589 Alan Hayward <alan.hayward@arm.com>
6590 David Sherwood <david.sherwood@arm.com>
6592 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
6594 (expand_expr_real_1): Update call accordingly.
6596 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6597 Alan Hayward <alan.hayward@arm.com>
6598 David Sherwood <david.sherwood@arm.com>
6600 * expmed.c (store_bit_field_using_insv): Add op0_mode and
6601 value_mode arguments. Use scalar_int_mode internally.
6602 (store_bit_field_1): Rename the new integer mode from imode
6603 to op0_mode and use it instead of GET_MODE (op0). Update calls
6604 to store_split_bit_field, store_bit_field_using_insv and
6605 store_fixed_bit_field.
6606 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
6607 Use scalar_int_mode internally. Use a bit count rather than a mode
6608 when calculating the largest bit size for get_best_mode.
6609 Update calls to store_split_bit_field and store_fixed_bit_field_1.
6610 (store_fixed_bit_field_1): Add mode and value_mode arguments.
6611 Remove assertion that OP0 has a scalar integer mode.
6612 (store_split_bit_field): Add op0_mode and value_mode arguments.
6613 Update calls to extract_fixed_bit_field.
6614 (extract_bit_field_using_extv): Add an op0_mode argument.
6615 Use scalar_int_mode internally.
6616 (extract_bit_field_1): Rename the new integer mode from imode to
6617 op0_mode and use it instead of GET_MODE (op0). Update calls to
6618 extract_split_bit_field, extract_bit_field_using_extv and
6619 extract_fixed_bit_field.
6620 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
6621 to extract_split_bit_field and extract_fixed_bit_field_1.
6622 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
6623 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
6625 (extract_split_bit_field): Add an op0_mode argument. Update call
6626 to extract_fixed_bit_field.
6628 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6629 Alan Hayward <alan.hayward@arm.com>
6630 David Sherwood <david.sherwood@arm.com>
6632 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
6633 * explow.c (hard_function_value): Likewise.
6634 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
6635 convert_to_mode call outside the loop.
6636 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
6637 for the mode iterator. Require the mode specified by max_pieces
6639 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
6641 (copy_blkmode_to_reg): Likewise.
6642 (set_storage_via_setmem): Likewise.
6643 * optabs.c (prepare_cmp_insn): Likewise.
6644 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
6645 * stor-layout.c (finish_bitfield_representative): Likewise.
6647 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6648 Alan Hayward <alan.hayward@arm.com>
6649 David Sherwood <david.sherwood@arm.com>
6651 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
6652 * expr.c (convert_move): Use them.
6653 (convert_modes): Likewise.
6654 (store_expr_with_bounds): Likewise.
6656 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6657 Alan Hayward <alan.hayward@arm.com>
6658 David Sherwood <david.sherwood@arm.com>
6660 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
6661 parameter for the mode of "x". Remove the "known_x", "known_mode"
6662 and "known_ret" arguments. Change the type of the mode argument
6664 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
6665 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
6666 (reg_num_sign_bit_copies_for_combine): Likewise.
6667 * rtlanal.c (nonzero_bits1): Likewise.
6668 (num_sign_bit_copies1): Likewise.
6669 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
6670 (reg_num_sign_bit_copies_general): Likewise.
6671 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
6672 (reg_nonzero_bits_general): Likewise.
6674 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6675 Alan Hayward <alan.hayward@arm.com>
6676 David Sherwood <david.sherwood@arm.com>
6678 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
6679 than in subroutines. Return 1 for non-integer modes.
6680 (cached_num_sign_bit_copies): Change the type of the mode parameter
6682 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
6683 classes. Handle CONST_INT_P first and then check whether X also
6684 has a scalar integer mode. Check the same thing for inner registers
6685 of a SUBREG and for values that are being extended or truncated.
6687 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6688 Alan Hayward <alan.hayward@arm.com>
6689 David Sherwood <david.sherwood@arm.com>
6691 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
6692 in subroutines. Return the mode mask for non-integer modes.
6693 (cached_nonzero_bits): Change the type of the mode parameter
6695 (nonzero_bits1): Likewise. Remove early exit for other mode
6696 classes. Handle CONST_INT_P first and then check whether X
6697 also has a scalar integer mode.
6699 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6700 Alan Hayward <alan.hayward@arm.com>
6701 David Sherwood <david.sherwood@arm.com>
6703 * expr.c (widest_int_mode_for_size): Make the comment match the code.
6704 Return a scalar_int_mode and assert that the size is greater than
6706 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
6707 (op_by_pieces_d::op_by_pieces_d): Likewise.
6708 (op_by_pieces_d::run): Likewise.
6709 (can_store_by_pieces): Likewise.
6711 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6712 Alan Hayward <alan.hayward@arm.com>
6713 David Sherwood <david.sherwood@arm.com>
6715 * combine.c (extract_left_shift): Add a mode argument and update
6717 (make_compound_operation_int): Change the type of the mode parameter
6718 to scalar_int_mode and update the call to extract_left_shift.
6720 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6721 Alan Hayward <alan.hayward@arm.com>
6722 David Sherwood <david.sherwood@arm.com>
6724 * combine.c (simplify_and_const_int): Change the type of the mode
6725 parameter to scalar_int_mode.
6726 (simplify_and_const_int_1): Likewise. Update recursive call.
6728 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6729 Alan Hayward <alan.hayward@arm.com>
6730 David Sherwood <david.sherwood@arm.com>
6732 * combine.c (simplify_compare_const): Check that the mode is a
6733 scalar_int_mode (rather than VOIDmode) before testing its
6735 (simplify_comparison): Move COMPARISON_P handling out of the
6736 loop and restrict the latter part of the loop to scalar_int_modes.
6737 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
6738 and when considering SUBREG_REGs. Use is_int_mode instead of
6739 checking GET_MODE_CLASS against MODE_INT.
6741 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6742 Alan Hayward <alan.hayward@arm.com>
6743 David Sherwood <david.sherwood@arm.com>
6745 * combine.c (try_widen_shift_mode): Move check for equal modes to...
6746 (simplify_shift_const_1): ...here. Use scalar_int_mode for
6747 shift_unit_mode and for modes involved in scalar shifts.
6749 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6750 Alan Hayward <alan.hayward@arm.com>
6751 David Sherwood <david.sherwood@arm.com>
6753 * combine.c (force_int_to_mode): New function, split out from...
6754 (force_to_mode): ...here. Keep xmode up-to-date and use it
6755 instead of GET_MODE (x).
6757 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6758 Alan Hayward <alan.hayward@arm.com>
6759 David Sherwood <david.sherwood@arm.com>
6761 * optabs-query.h (extraction_insn::struct_mode): Change type to
6762 opt_scalar_int_mode and update comment.
6763 (extraction_insn::field_mode): Change type to scalar_int_mode.
6764 (extraction_insn::pos_mode): Likewise.
6765 * combine.c (make_extraction): Update accordingly.
6766 * optabs-query.c (get_traditional_extraction_insn): Likewise.
6767 (get_optab_extraction_insn): Likewise.
6768 * recog.c (simplify_while_replacing): Likewise.
6769 * expmed.c (narrow_bit_field_mem): Change the type of the mode
6770 parameter to opt_scalar_int_mode.
6772 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6773 Alan Hayward <alan.hayward@arm.com>
6774 David Sherwood <david.sherwood@arm.com>
6776 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
6777 to a scalar_int_mode instead of a machine_mode.
6778 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
6779 (get_best_mode): Return a boolean and use a pointer argument to store
6780 the selected mode. Replace the limit mode parameter with a bit limit.
6781 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
6782 for the values returned by bit_field_mode_iterator::next_mode.
6783 (store_bit_field): Update call to get_best_mode.
6784 (store_fixed_bit_field): Likewise.
6785 (extract_fixed_bit_field): Likewise.
6786 * expr.c (optimize_bitfield_assignment_op): Likewise.
6787 * fold-const.c (optimize_bit_field_compare): Likewise.
6788 (fold_truth_andor_1): Likewise.
6789 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
6790 Update for new type of m_mode.
6791 (get_best_mode): As above.
6793 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6794 Alan Hayward <alan.hayward@arm.com>
6795 David Sherwood <david.sherwood@arm.com>
6797 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
6798 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
6799 (store_bit_field): Check is_a <scalar_int_mode> before calling
6800 strict_volatile_bitfield_p.
6801 (extract_bit_field): Likewise.
6803 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6804 Alan Hayward <alan.hayward@arm.com>
6805 David Sherwood <david.sherwood@arm.com>
6807 * target.def (cstore_mode): Return a scalar_int_mode.
6808 * doc/tm.texi: Regenerate.
6809 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
6810 * targhooks.h (default_cstore_mode): Likewise.
6811 * targhooks.c (default_cstore_mode): Likewise, using a forced
6813 * expmed.c (emit_cstore): Expect the target of the cstore to be
6816 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6817 Alan Hayward <alan.hayward@arm.com>
6818 David Sherwood <david.sherwood@arm.com>
6820 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
6822 (niter_desc): Likewise mode.
6823 (iv_analyze): Add a mode parameter.
6825 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
6826 and change its type to scalar_int_mode.
6827 * loop-iv.c: Update commentary at head of file.
6828 (iv_constant): Pass the mode paraeter before the rtx it describes
6829 and change its type to scalar_int_mode. Remove VOIDmode handling.
6830 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
6831 (iv_extend): Likewise.
6832 (shorten_into_mode): Likewise.
6833 (iv_add): Use scalar_int_mode.
6834 (iv_mult): Likewise.
6835 (iv_shift): Likewise.
6836 (canonicalize_iv_subregs): Likewise.
6837 (get_biv_step_1): Pass the outer_mode parameter before the rtx
6838 it describes and change its mode to scalar_int_mode. Also change
6839 the type of the returned inner_mode to scalar_int_mode.
6840 (get_biv_step): Likewise, turning outer_mode from a pointer
6841 into a direct parameter. Update call to get_biv_step_1.
6842 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
6843 iv_constant and get_biv_step.
6844 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
6845 and change its type to scalar_int_mode. Don't initialise iv->mode
6846 to VOIDmode and remove later checks for its still being VOIDmode.
6847 Update calls to iv_analyze_op and iv_analyze_expr. Check
6848 is_a <scalar_int_mode> when changing the mode under consideration.
6849 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
6850 Update call to iv_analyze_expr.
6851 (iv_analyze_op): Add a mode parameter. Reject subregs whose
6852 inner register is not also a scalar_int_mode. Update call to
6854 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
6855 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
6856 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
6857 separate mode class checks. Update calls to iv_analyze. Remove
6858 fix-up of VOIDmodes after iv_analyze_biv.
6859 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
6860 don't have a scalar_int_mode. Update call to biv_p.
6862 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6863 Alan Hayward <alan.hayward@arm.com>
6864 David Sherwood <david.sherwood@arm.com>
6866 * cfgexpand.c (convert_debug_memory_address): Use
6867 as_a <scalar_int_mode>.
6868 * combine.c (expand_compound_operation): Likewise.
6869 (make_extraction): Likewise.
6870 (change_zero_ext): Likewise.
6871 (simplify_comparison): Likewise.
6872 * cse.c (cse_insn): Likewise.
6873 * dwarf2out.c (minmax_loc_descriptor): Likewise.
6874 (mem_loc_descriptor): Likewise.
6875 (loc_descriptor): Likewise.
6876 * expmed.c (init_expmed_one_mode): Likewise.
6877 (synth_mult): Likewise.
6878 (emit_store_flag_1): Likewise.
6879 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
6880 of a comparison with size.
6881 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
6882 (reduce_to_bit_field_precision): Likewise.
6883 * function.c (expand_function_end): Likewise.
6884 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
6885 * loop-doloop.c (doloop_modify): Likewise.
6886 * optabs.c (expand_binop): Likewise.
6887 (expand_unop): Likewise.
6888 (expand_copysign_absneg): Likewise.
6889 (prepare_cmp_insn): Likewise.
6890 (maybe_legitimize_operand): Likewise.
6891 * recog.c (const_scalar_int_operand): Likewise.
6892 * rtlanal.c (get_address_mode): Likewise.
6893 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6894 (simplify_cond_clz_ctz): Likewise.
6895 * tree-nested.c (get_nl_goto_field): Likewise.
6896 * tree.c (build_vector_type_for_mode): Likewise.
6897 * var-tracking.c (use_narrower_mode): Likewise.
6899 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6900 Alan Hayward <alan.hayward@arm.com>
6901 David Sherwood <david.sherwood@arm.com>
6903 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
6904 * builtins.c (expand_builtin_signbit): Use it.
6905 * cfgexpand.c (expand_debug_expr): Likewise.
6906 * dojump.c (do_jump): Likewise.
6907 (do_compare_and_jump): Likewise.
6908 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
6909 * expmed.c (make_tree): Likewise.
6910 * expr.c (expand_expr_real_2): Likewise.
6911 (expand_expr_real_1): Likewise.
6912 (try_casesi): Likewise.
6913 * fold-const-call.c (fold_const_call_ss): Likewise.
6914 * fold-const.c (unextend): Likewise.
6915 (extract_muldiv_1): Likewise.
6916 (fold_single_bit_test): Likewise.
6917 (native_encode_int): Likewise.
6918 (native_encode_string): Likewise.
6919 (native_interpret_int): Likewise.
6920 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
6921 * internal-fn.c (expand_addsub_overflow): Likewise.
6922 (expand_neg_overflow): Likewise.
6923 (expand_mul_overflow): Likewise.
6924 (expand_arith_overflow): Likewise.
6925 * match.pd: Likewise.
6926 * stor-layout.c (layout_type): Likewise.
6927 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
6928 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
6929 * tree-ssanames.c (get_range_info): Likewise.
6930 * tree-switch-conversion.c (array_value_type) Likewise.
6931 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
6932 (vect_recog_divmod_pattern): Likewise.
6933 (vect_recog_mixed_size_cond_pattern): Likewise.
6934 * tree-vrp.c (extract_range_basic): Likewise.
6935 (simplify_float_conversion_using_ranges): Likewise.
6936 * tree.c (int_fits_type_p): Likewise.
6937 * ubsan.c (instrument_bool_enum_load): Likewise.
6938 * varasm.c (mergeable_string_section): Likewise.
6939 (narrowing_initializer_constant_valid_p): Likewise.
6940 (output_constant): Likewise.
6942 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6943 Alan Hayward <alan.hayward@arm.com>
6944 David Sherwood <david.sherwood@arm.com>
6946 * machmode.h (NARROWEST_INT_MODE): New macro.
6947 * expr.c (alignment_for_piecewise_move): Use it instead of
6948 GET_CLASS_NARROWEST_MODE (MODE_INT).
6949 (push_block): Likewise.
6950 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
6952 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6954 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6955 Alan Hayward <alan.hayward@arm.com>
6956 David Sherwood <david.sherwood@arm.com>
6958 * postreload.c (move2add_valid_value_p): Change the type of the
6959 mode parameter to scalar_int_mode.
6960 (move2add_use_add2_insn): Add a mode parameter and use it instead
6962 (move2add_use_add3_insn): Likewise.
6963 (reload_cse_move2add): Update accordingly.
6965 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6966 Alan Hayward <alan.hayward@arm.com>
6967 David Sherwood <david.sherwood@arm.com>
6969 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
6971 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
6972 * optabs.c (expand_unop): Likewise.
6974 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6975 Alan Hayward <alan.hayward@arm.com>
6976 David Sherwood <david.sherwood@arm.com>
6978 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
6979 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
6980 (popcount_loc_descriptor): Likewise.
6981 (bswap_loc_descriptor): Likewise.
6982 (rotate_loc_descriptor): Likewise.
6983 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
6984 calling the functions above.
6986 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6987 Alan Hayward <alan.hayward@arm.com>
6988 David Sherwood <david.sherwood@arm.com>
6990 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
6992 (try_combine): Likewise.
6993 (simplify_if_then_else): Likewise.
6994 * cse.c (cse_insn): Likewise.
6995 * dwarf2out.c (mem_loc_descriptor): Likewise.
6996 * emit-rtl.c (gen_lowpart_common): Likewise.
6997 * simplify-rtx.c (simplify_truncation): Likewise.
6998 (simplify_binary_operation_1): Likewise.
6999 (simplify_const_relational_operation): Likewise.
7000 (simplify_ternary_operation): Likewise.
7001 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
7003 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7004 Alan Hayward <alan.hayward@arm.com>
7005 David Sherwood <david.sherwood@arm.com>
7007 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
7008 * reload.c (push_reload): Likewise.
7009 (find_reloads): Likewise.
7011 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7012 Alan Hayward <alan.hayward@arm.com>
7013 David Sherwood <david.sherwood@arm.com>
7015 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
7016 (make_compound_operation_int): Likewise.
7017 (change_zero_ext): Likewise.
7018 * expr.c (convert_move): Likewise.
7019 (convert_modes): Likewise.
7020 * fwprop.c (forward_propagate_subreg): Likewise.
7021 * loop-iv.c (get_biv_step_1): Likewise.
7022 * optabs.c (widen_operand): Likewise.
7023 * postreload.c (move2add_valid_value_p): Likewise.
7024 * recog.c (simplify_while_replacing): Likewise.
7025 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7026 (simplify_binary_operation_1): Likewise. Remove redundant
7027 mode equality check.
7029 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7030 Alan Hayward <alan.hayward@arm.com>
7031 David Sherwood <david.sherwood@arm.com>
7033 * combine.c (combine_simplify_rtx): Add checks for
7034 is_a <scalar_int_mode>.
7035 (simplify_if_then_else): Likewise.
7036 (make_field_assignment): Likewise.
7037 (simplify_comparison): Likewise.
7038 * ifcvt.c (noce_try_bitop): Likewise.
7039 * loop-invariant.c (canonicalize_address_mult): Likewise.
7040 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7042 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7043 Alan Hayward <alan.hayward@arm.com>
7044 David Sherwood <david.sherwood@arm.com>
7046 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
7047 is_a <scalar_int_mode> instead of != BLKmode.
7049 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7050 Alan Hayward <alan.hayward@arm.com>
7051 David Sherwood <david.sherwood@arm.com>
7053 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
7054 instead of != VOIDmode.
7055 * combine.c (if_then_else_cond): Likewise.
7056 (change_zero_ext): Likewise.
7057 * dwarf2out.c (mem_loc_descriptor): Likewise.
7058 (loc_descriptor): Likewise.
7059 * rtlanal.c (canonicalize_condition): Likewise.
7060 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
7062 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7063 Alan Hayward <alan.hayward@arm.com>
7064 David Sherwood <david.sherwood@arm.com>
7066 * simplify-rtx.c (simplify_binary_operation_1): Use
7067 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
7069 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7070 Alan Hayward <alan.hayward@arm.com>
7071 David Sherwood <david.sherwood@arm.com>
7073 * wide-int.h (int_traits<unsigned char>) New class.
7074 (int_traits<unsigned short>) Likewise.
7075 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
7076 Use GET_MODE_UNIT_PRECISION and remove redundant test for
7078 * combine.c (set_nonzero_bits_and_sign_copies): Use
7079 is_a <scalar_int_mode>.
7080 (find_split_point): Likewise.
7081 (combine_simplify_rtx): Likewise.
7082 (simplify_logical): Likewise.
7083 (expand_compound_operation): Likewise.
7084 (expand_field_assignment): Likewise.
7085 (make_compound_operation): Likewise.
7086 (extended_count): Likewise.
7087 (change_zero_ext): Likewise.
7088 (simplify_comparison): Likewise.
7089 * dwarf2out.c (scompare_loc_descriptor): Likewise.
7090 (ucompare_loc_descriptor): Likewise.
7091 (minmax_loc_descriptor): Likewise.
7092 (mem_loc_descriptor): Likewise.
7093 (loc_descriptor): Likewise.
7094 * expmed.c (init_expmed_one_mode): Likewise.
7095 * lra-constraints.c (lra_constraint_offset): Likewise.
7096 * optabs.c (prepare_libcall_arg): Likewise.
7097 * postreload.c (move2add_note_store): Likewise.
7098 * reload.c (operands_match_p): Likewise.
7099 * rtl.h (load_extend_op): Likewise.
7100 * rtlhooks.c (gen_lowpart_general): Likewise.
7101 * simplify-rtx.c (simplify_truncation): Likewise.
7102 (simplify_unary_operation_1): Likewise.
7103 (simplify_binary_operation_1): Likewise.
7104 (simplify_const_binary_operation): Likewise.
7105 (simplify_const_relational_operation): Likewise.
7106 (simplify_subreg): Likewise.
7107 * stor-layout.c (bitwise_mode_for_mode): Likewise.
7108 * var-tracking.c (adjust_mems): Likewise.
7109 (prepare_call_arguments): Likewise.
7111 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7112 Alan Hayward <alan.hayward@arm.com>
7113 David Sherwood <david.sherwood@arm.com>
7115 * machmode.h (is_int_mode): New fuction.
7116 * combine.c (find_split_point): Use it.
7117 (combine_simplify_rtx): Likewise.
7118 (simplify_if_then_else): Likewise.
7119 (simplify_set): Likewise.
7120 (simplify_shift_const_1): Likewise.
7121 (simplify_comparison): Likewise.
7122 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
7123 * cse.c (notreg_cost): Likewise.
7124 (cse_insn): Likewise.
7125 * cselib.c (cselib_lookup_1): Likewise.
7126 * dojump.c (do_jump_1): Likewise.
7127 (do_compare_rtx_and_jump): Likewise.
7128 * dse.c (get_call_args): Likewise.
7129 * dwarf2out.c (rtl_for_decl_init): Likewise.
7130 (native_encode_initializer): Likewise.
7131 * expmed.c (emit_store_flag_1): Likewise.
7132 (emit_store_flag): Likewise.
7133 * expr.c (convert_modes): Likewise.
7134 (store_field): Likewise.
7135 (expand_expr_real_1): Likewise.
7136 * fold-const.c (fold_read_from_constant_string): Likewise.
7137 * gimple-ssa-sprintf.c (get_format_string): Likewise.
7138 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
7139 * optabs.c (expand_binop): Likewise.
7140 (expand_unop): Likewise.
7141 (expand_abs_nojump): Likewise.
7142 (expand_one_cmpl_abs_nojump): Likewise.
7143 * simplify-rtx.c (mode_signbit_p): Likewise.
7144 (val_signbit_p): Likewise.
7145 (val_signbit_known_set_p): Likewise.
7146 (val_signbit_known_clear_p): Likewise.
7147 (simplify_relational_operation_1): Likewise.
7148 * tree.c (vector_type_mode): Likewise.
7150 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7151 Alan Hayward <alan.hayward@arm.com>
7152 David Sherwood <david.sherwood@arm.com>
7154 * machmode.h (smallest_mode_for_size): Fix formatting.
7155 (smallest_int_mode_for_size): New function.
7156 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
7157 instead of smallest_mode_for_size.
7158 * combine.c (make_extraction): Likewise.
7159 * config/arc/arc.c (arc_expand_movmem): Likewise.
7160 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
7161 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
7162 * config/s390/s390.c (s390_expand_insv): Likewise.
7163 * config/sparc/sparc.c (assign_int_registers): Likewise.
7164 * config/spu/spu.c (spu_function_value): Likewise.
7165 (spu_function_arg): Likewise.
7166 * coverage.c (get_gcov_type): Likewise.
7167 (get_gcov_unsigned_t): Likewise.
7168 * dse.c (find_shift_sequence): Likewise.
7169 * expmed.c (store_bit_field_1): Likewise.
7170 * expr.c (convert_move): Likewise.
7171 (store_field): Likewise.
7172 * internal-fn.c (expand_arith_overflow): Likewise.
7173 * optabs-query.c (get_best_extraction_insn): Likewise.
7174 * optabs.c (expand_twoval_binop_libfunc): Likewise.
7175 * stor-layout.c (layout_type): Likewise.
7176 (initialize_sizetypes): Likewise.
7177 * targhooks.c (default_get_mask_mode): Likewise.
7178 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
7180 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7181 Alan Hayward <alan.hayward@arm.com>
7182 David Sherwood <david.sherwood@arm.com>
7184 * machmode.h (opt_mode::else_blk): New function.
7185 (int_mode_for_mode): Declare.
7186 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
7187 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
7189 * cfgexpand.c (expand_debug_expr): Likewise.
7190 * combine.c (gen_lowpart_or_truncate): Likewise.
7191 (gen_lowpart_for_combine): Likewise.
7192 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
7193 * config/avr/avr.c (avr_to_int_mode): Likewise.
7194 (avr_out_plus_1): Likewise.
7195 (avr_out_plus): Likewise.
7196 (avr_out_round): Likewise.
7197 * config/i386/i386.c (ix86_split_to_parts): Likewise.
7198 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
7199 (s390_expand_vcond): Likewise.
7200 * config/spu/spu.c (spu_split_immediate): Likewise.
7201 (spu_expand_mov): Likewise.
7202 * dse.c (get_stored_val): Likewise.
7203 * expmed.c (store_bit_field_1): Likewise.
7204 (convert_extracted_bit_field): Use int_mode_for_mode instead of
7206 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
7207 (extract_low_bits): Likewise.
7208 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
7209 handling rather than repeating the check.
7210 (emit_group_store): Likewise.
7211 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
7212 * optabs.c (expand_absneg_bit): Likewise.
7213 (expand_copysign_absneg): Likewise.
7214 (expand_copysign_bit): Likewise.
7215 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
7216 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
7217 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
7218 * var-tracking.c (prepare_call_arguments): Likewise.
7219 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
7220 int_mode_for_mode instead of mode_for_size.
7221 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
7223 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7224 Alan Hayward <alan.hayward@arm.com>
7225 David Sherwood <david.sherwood@arm.com>
7227 * machmode.h (int_mode_for_size): New function.
7228 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
7229 instead of mode_for_size.
7230 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
7232 * combine.c (expand_field_assignment): Use int_mode_for_size
7233 instead of mode_for_size.
7234 (make_extraction): Likewise.
7235 (simplify_shift_const_1): Likewise.
7236 (simplify_comparison): Likewise.
7237 * dojump.c (do_jump): Likewise.
7238 * dwarf2out.c (mem_loc_descriptor): Likewise.
7239 * emit-rtl.c (init_derived_machine_modes): Likewise.
7240 * expmed.c (flip_storage_order): Likewise.
7241 (convert_extracted_bit_field): Likewise.
7242 * expr.c (copy_blkmode_from_reg): Likewise.
7243 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
7244 * internal-fn.c (expand_mul_overflow): Likewise.
7245 * lower-subreg.c (simple_move): Likewise.
7246 * optabs-libfuncs.c (init_optabs): Likewise.
7247 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7248 * tree.c (vector_type_mode): Likewise.
7249 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
7250 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
7251 * tree-vect-generic.c (expand_vector_parallel): Likewise.
7252 * tree-vect-stmts.c (vectorizable_load): Likewise.
7253 (vectorizable_store): Likewise.
7255 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7256 Alan Hayward <alan.hayward@arm.com>
7257 David Sherwood <david.sherwood@arm.com>
7259 * coretypes.h (pod_mode): New type.
7260 (scalar_int_mode_pod): New typedef.
7261 * machmode.h (pod_mode): New class.
7262 (int_n_data_t::m): Change type to scalar_int_mode_pod.
7263 * genmodes.c (emit_mode_int_n): Update accordingly.
7264 * lower-subreg.h (target_lower_subreg): Change type to
7265 scalar_int_mode_pod.
7266 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
7267 scalar_int_mode_pod.
7269 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7270 Alan Hayward <alan.hayward@arm.com>
7271 David Sherwood <david.sherwood@arm.com>
7273 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
7274 machine_mode to scalar_int_mode.
7275 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
7276 (rs6000_option_override_internal): Remove cast to int.
7277 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
7278 machine_mode to scalar_int_mode.
7279 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
7280 (rs6000_option_override_internal): Remove cast to int.
7281 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
7282 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
7284 * config/s390/s390.c (s390_expand_builtin): Likewise.
7285 * coretypes.h (scalar_int_mode): New type.
7286 (opt_scalar_int_mode): New typedef.
7287 * machmode.h (scalar_int_mode): New class.
7288 (scalar_int_mode::includes_p): New function.
7289 (byte_mode): Change type to scalar_int_mode.
7290 (word_mode): Likewise.
7291 (ptr_mode): Likewise.
7292 * emit-rtl.c (byte_mode): Likewise.
7293 (word_mode): Likewise.
7294 (ptr_mode): Likewise.
7295 (init_derived_machine_modes): Update accordingly.
7296 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
7297 and MODE_PARTIAL_INT.
7298 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
7299 opt_scalar_int_mode.
7301 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7302 Alan Hayward <alan.hayward@arm.com>
7303 David Sherwood <david.sherwood@arm.com>
7305 * target.def (libgcc_floating_mode_supported_p): Take a
7307 * doc/tm.texi: Regenerate.
7308 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
7310 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
7311 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
7314 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7315 Alan Hayward <alan.hayward@arm.com>
7316 David Sherwood <david.sherwood@arm.com>
7318 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
7319 * doc/tm.texi: Regenerate.
7320 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
7321 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
7322 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
7323 * targhooks.h (default_floatn_mode): Likewise.
7324 * targhooks.c (default_floatn_mode): Likewise.
7325 * tree.c (build_common_tree_nodes): Update accordingly.
7327 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7328 Alan Hayward <alan.hayward@arm.com>
7329 David Sherwood <david.sherwood@arm.com>
7331 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
7332 (mode_iterator::iterate_p): Likewise.
7333 (mode_iterator::get_wider): Likewise.
7334 * expr.c (init_expr_target): Use opt_scalar_float_mode.
7336 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7337 Alan Hayward <alan.hayward@arm.com>
7338 David Sherwood <david.sherwood@arm.com>
7340 * coretypes.h (opt_scalar_float_mode): New typedef.
7341 * machmode.h (float_mode_for_size): New function.
7342 * emit-rtl.c (double_mode): Delete.
7343 (init_emit_once): Use float_mode_for_size.
7344 * stor-layout.c (layout_type): Likewise.
7345 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
7347 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7348 Alan Hayward <alan.hayward@arm.com>
7349 David Sherwood <david.sherwood@arm.com>
7351 * output.h (assemble_real): Take a scalar_float_mode.
7352 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
7353 * config/arm/arm.md (consttable_4): Likewise.
7354 (consttable_8): Likewise.
7355 (consttable_16): Likewise.
7356 * config/mips/mips.md (consttable_float): Likewise.
7357 * config/s390/s390.c (s390_output_pool_entry): Likewise.
7358 * varasm.c (assemble_real): Take a scalar_float_mode.
7359 (output_constant_pool_2): Update accordingly.
7360 (output_constant): Likewise.
7362 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7363 Alan Hayward <alan.hayward@arm.com>
7364 David Sherwood <david.sherwood@arm.com>
7366 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
7367 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
7368 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
7369 (native_encode_real): Likewise.
7370 (native_interpret_real): Likewise.
7371 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
7372 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7374 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7375 Alan Hayward <alan.hayward@arm.com>
7376 David Sherwood <david.sherwood@arm.com>
7378 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
7379 <scalar_float_mode>. Simplify.
7380 (gen_extend_conv_libfunc): Likewise.
7382 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7383 Alan Hayward <alan.hayward@arm.com>
7384 David Sherwood <david.sherwood@arm.com>
7386 * coretypes.h (scalar_float_mode): New type.
7387 * machmode.h (mode_traits::from_int): Use machine_mode if
7388 USE_ENUM_MODES is defined.
7389 (is_a): New function.
7391 (dyn_cast): Likewise.
7392 (scalar_float_mode): New class.
7393 (scalar_float_mode::includes_p): New function.
7394 (is_float_mode): Likewise.
7395 * gdbhooks.py (MachineModePrinter): New class.
7396 (build_pretty_printer): Use it for scalar_float_mode.
7397 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
7398 (format_helper::format_helper): Turn into a template.
7399 * genmodes.c (get_mode_class): New function.
7400 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
7401 or machine_mode if none.
7402 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
7403 as_a <scalar_float_mode>.
7404 * dwarf2out.c (mem_loc_descriptor): Likewise.
7405 (insert_float): Likewise.
7406 (add_const_value_attribute): Likewise.
7407 * simplify-rtx.c (simplify_immed_subreg): Likewise.
7408 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
7409 (expand_unop): Update accordingly.
7410 (expand_abs_nojump): Likewise.
7411 (expand_copysign_absneg): Take a scalar_float_mode.
7412 (expand_copysign_bit): Likewise.
7413 (expand_copysign): Update accordingly.
7415 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7416 Alan Hayward <alan.hayward@arm.com>
7417 David Sherwood <david.sherwood@arm.com>
7419 * coretypes.h (opt_mode): New class.
7420 * machmode.h (opt_mode): Likewise.
7421 (opt_mode::else_void): New function.
7422 (opt_mode::require): Likewise.
7423 (opt_mode::exists): Likewise.
7424 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
7425 (GET_MODE_2XWIDER_MODE): Likewise.
7426 (mode_iterator::get_wider): Update accordingly.
7427 (mode_iterator::get_2xwider): Likewise.
7428 (mode_iterator::get_known_wider): Likewise, turning into a template.
7429 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
7430 forcing a wider mode to exist.
7431 * config/cr16/cr16.h (LONG_REG_P): Likewise.
7432 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
7433 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
7434 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
7435 * lower-subreg.c (init_lower_subreg): Likewise.
7436 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
7437 on the final iteration.
7438 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
7439 a wider mode exists before asking for a move pattern.
7440 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
7441 forcing a wider mode to exist.
7442 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
7443 returning false if no such mode exists.
7444 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
7445 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
7446 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
7447 Avoid checking for a MODE_INT if we already know the mode is not a
7449 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
7450 forcing a wider mode to exist.
7451 (expmed_mult_highpart_optab): Likewise.
7452 (expmed_mult_highpart): Likewise.
7453 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
7455 * lto-streamer-in.c (lto_input_mode_table): Likewise.
7456 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
7457 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
7458 * internal-fn.c (expand_mul_overflow): Update use of
7459 GET_MODE_2XWIDER_MODE.
7460 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7461 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
7462 GET_MODE_WIDER_MODE.
7463 (convert_plusminus_to_widen): Likewise.
7464 * tree-switch-conversion.c (array_value_type): Likewise.
7465 * var-tracking.c (emit_note_insn_var_location): Likewise.
7466 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7467 Return false inside rather than outside the loop if no wider mode
7469 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
7470 and GET_MODE_2XWIDER_MODE
7471 (can_compare_p): Use else_void.
7472 * gdbhooks.py (OptMachineModePrinter): New class.
7473 (build_pretty_printer): Use it for opt_mode.
7475 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7476 Alan Hayward <alan.hayward@arm.com>
7477 David Sherwood <david.sherwood@arm.com>
7479 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
7480 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
7482 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7483 Alan Hayward <alan.hayward@arm.com>
7484 David Sherwood <david.sherwood@arm.com>
7486 * machmode.h (mode_traits): New structure.
7487 (get_narrowest_mode): New function.
7488 (mode_iterator::start): Likewise.
7489 (mode_iterator::iterate_p): Likewise.
7490 (mode_iterator::get_wider): Likewise.
7491 (mode_iterator::get_known_wider): Likewise.
7492 (mode_iterator::get_2xwider): Likewise.
7493 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
7494 (FOR_EACH_MODE): Likewise.
7495 (FOR_EACH_MODE_FROM): Likewise.
7496 (FOR_EACH_MODE_UNTIL): Likewise.
7497 (FOR_EACH_WIDER_MODE): Likewise.
7498 (FOR_EACH_2XWIDER_MODE): Likewise.
7499 * builtins.c (expand_builtin_strlen): Use new mode iterators.
7500 * combine.c (simplify_comparison): Likewise
7501 * config/i386/i386.c (type_natural_mode): Likewise.
7502 * cse.c (cse_insn): Likewise.
7503 * dse.c (find_shift_sequence): Likewise.
7504 * emit-rtl.c (init_derived_machine_modes): Likewise.
7505 (init_emit_once): Likewise.
7506 * explow.c (hard_function_value): Likewise.
7507 * expmed.c (extract_fixed_bit_field_1): Likewise.
7508 (extract_bit_field_1): Likewise.
7509 (expand_divmod): Likewise.
7510 (emit_store_flag_1): Likewise.
7511 * expr.c (init_expr_target): Likewise.
7512 (convert_move): Likewise.
7513 (alignment_for_piecewise_move): Likewise.
7514 (widest_int_mode_for_size): Likewise.
7515 (emit_block_move_via_movmem): Likewise.
7516 (copy_blkmode_to_reg): Likewise.
7517 (set_storage_via_setmem): Likewise.
7518 (compress_float_constant): Likewise.
7519 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7520 * optabs-query.c (get_best_extraction_insn): Likewise.
7521 * optabs.c (expand_binop): Likewise.
7522 (expand_twoval_unop): Likewise.
7523 (expand_twoval_binop): Likewise.
7524 (widen_leading): Likewise.
7525 (widen_bswap): Likewise.
7526 (expand_parity): Likewise.
7527 (expand_unop): Likewise.
7528 (prepare_cmp_insn): Likewise.
7529 (prepare_float_lib_cmp): Likewise.
7530 (expand_float): Likewise.
7531 (expand_fix): Likewise.
7532 (expand_sfix_optab): Likewise.
7533 * postreload.c (move2add_use_add2_insn): Likewise.
7534 * reg-stack.c (reg_to_stack): Likewise.
7535 * reginfo.c (choose_hard_reg_mode): Likewise.
7536 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
7537 * stor-layout.c (mode_for_size): Likewise.
7538 (smallest_mode_for_size): Likewise.
7539 (mode_for_vector): Likewise.
7540 (finish_bitfield_representative): Likewise.
7541 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
7542 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
7543 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7544 * var-tracking.c (prepare_call_arguments): Likewise.
7546 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7547 Alan Hayward <alan.hayward@arm.com>
7548 David Sherwood <david.sherwood@arm.com>
7550 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
7551 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
7552 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
7553 * machmode.h (mode_size): Move earlier in file.
7554 (mode_precision): Likewise.
7555 (mode_inner): Likewise.
7556 (mode_nunits): Likewise.
7557 (mode_unit_size): Likewise.
7558 (unit_unit_precision): Likewise.
7559 (mode_wider): Likewise.
7560 (mode_2xwider): Likewise.
7561 (machine_mode): New class.
7562 (mode_to_bytes): New function.
7563 (mode_to_bits): Likewise.
7564 (mode_to_precision): Likewise.
7565 (mode_to_inner): Likewise.
7566 (mode_to_unit_size): Likewise.
7567 (mode_to_unit_precision): Likewise.
7568 (mode_to_nunits): Likewise.
7569 (GET_MODE_SIZE): Use mode_to_bytes.
7570 (GET_MODE_BITSIZE): Use mode_to_bits.
7571 (GET_MODE_PRECISION): Use mode_to_precision.
7572 (GET_MODE_INNER): Use mode_to_inner.
7573 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
7574 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
7575 (GET_MODE_NUNITS): Use mode_to_nunits.
7576 * system.h (ALWAYS_INLINE): New macro.
7577 * config/powerpcspe/powerpcspe-c.c
7578 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
7579 int for arg1_mode and arg2_mode.
7581 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7582 Alan Hayward <alan.hayward@arm.com>
7583 David Sherwood <david.sherwood@arm.com>
7585 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
7586 Prefix mode names with E_ in case statements.
7587 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7588 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
7589 (aarch64_split_simd_move): Likewise.
7590 (aarch64_gen_storewb_pair): Likewise.
7591 (aarch64_gen_loadwb_pair): Likewise.
7592 (aarch64_gen_store_pair): Likewise.
7593 (aarch64_gen_load_pair): Likewise.
7594 (aarch64_get_condition_code_1): Likewise.
7595 (aarch64_constant_pool_reload_icode): Likewise.
7596 (get_rsqrte_type): Likewise.
7597 (get_rsqrts_type): Likewise.
7598 (get_recpe_type): Likewise.
7599 (get_recps_type): Likewise.
7600 (aarch64_gimplify_va_arg_expr): Likewise.
7601 (aarch64_simd_container_mode): Likewise.
7602 (aarch64_emit_load_exclusive): Likewise.
7603 (aarch64_emit_store_exclusive): Likewise.
7604 (aarch64_expand_compare_and_swap): Likewise.
7605 (aarch64_gen_atomic_cas): Likewise.
7606 (aarch64_emit_bic): Likewise.
7607 (aarch64_emit_atomic_swap): Likewise.
7608 (aarch64_emit_atomic_load_op): Likewise.
7609 (aarch64_evpc_trn): Likewise.
7610 (aarch64_evpc_uzp): Likewise.
7611 (aarch64_evpc_zip): Likewise.
7612 (aarch64_evpc_ext): Likewise.
7613 (aarch64_evpc_rev): Likewise.
7614 (aarch64_evpc_dup): Likewise.
7615 (aarch64_gen_ccmp_first): Likewise.
7616 (aarch64_gen_ccmp_next): Likewise.
7617 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
7618 (alpha_emit_xfloating_libcall): Likewise.
7619 (emit_insxl): Likewise.
7620 (alpha_arg_type): Likewise.
7621 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
7622 (arc_preferred_simd_mode): Likewise.
7623 (arc_secondary_reload): Likewise.
7624 (get_arc_condition_code): Likewise.
7625 (arc_print_operand): Likewise.
7626 (arc_legitimate_constant_p): Likewise.
7627 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7628 * config/arc/arc.md (casesi_load): Likewise.
7629 (casesi_compact_jump): Likewise.
7630 * config/arc/predicates.md (proper_comparison_operator): Likewise.
7631 (cc_use_register): Likewise.
7632 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7633 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
7634 (arm_init_iwmmxt_builtins): Likewise.
7635 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
7636 (neon_expand_vector_init): Likewise.
7637 (arm_attr_length_move_neon): Likewise.
7638 (maybe_get_arm_condition_code): Likewise.
7639 (arm_emit_vector_const): Likewise.
7640 (arm_preferred_simd_mode): Likewise.
7641 (arm_output_iwmmxt_tinsr): Likewise.
7642 (thumb1_output_casesi): Likewise.
7643 (thumb2_output_casesi): Likewise.
7644 (arm_emit_load_exclusive): Likewise.
7645 (arm_emit_store_exclusive): Likewise.
7646 (arm_expand_compare_and_swap): Likewise.
7647 (arm_evpc_neon_vuzp): Likewise.
7648 (arm_evpc_neon_vzip): Likewise.
7649 (arm_evpc_neon_vrev): Likewise.
7650 (arm_evpc_neon_vtrn): Likewise.
7651 (arm_evpc_neon_vext): Likewise.
7652 (arm_validize_comparison): Likewise.
7653 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
7654 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
7655 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
7656 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
7657 (c6x_preferred_simd_mode): Likewise.
7658 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
7659 (epiphany_rtx_costs): Likewise.
7660 * config/epiphany/predicates.md (proper_comparison_operator):
7662 * config/frv/frv.c (condexec_memory_operand): Likewise.
7663 (frv_emit_move): Likewise.
7664 (output_move_single): Likewise.
7665 (output_condmove_single): Likewise.
7666 (frv_hard_regno_mode_ok): Likewise.
7667 (frv_matching_accg_mode): Likewise.
7668 * config/h8300/h8300.c (split_adds_subs): Likewise.
7669 (h8300_rtx_costs): Likewise.
7670 (h8300_print_operand): Likewise.
7671 (compute_mov_length): Likewise.
7672 (output_logical_op): Likewise.
7673 (compute_logical_op_length): Likewise.
7674 (compute_logical_op_cc): Likewise.
7675 (h8300_shift_needs_scratch_p): Likewise.
7676 (output_a_shift): Likewise.
7677 (compute_a_shift_length): Likewise.
7678 (compute_a_shift_cc): Likewise.
7679 (expand_a_rotate): Likewise.
7680 (output_a_rotate): Likewise.
7681 * config/i386/i386.c (classify_argument): Likewise.
7682 (function_arg_advance_32): Likewise.
7683 (function_arg_32): Likewise.
7684 (function_arg_64): Likewise.
7685 (function_value_64): Likewise.
7686 (ix86_gimplify_va_arg): Likewise.
7687 (ix86_legitimate_constant_p): Likewise.
7688 (put_condition_code): Likewise.
7689 (split_double_mode): Likewise.
7690 (ix86_avx256_split_vector_move_misalign): Likewise.
7691 (ix86_expand_vector_logical_operator): Likewise.
7692 (ix86_split_idivmod): Likewise.
7693 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
7694 (ix86_build_const_vector): Likewise.
7695 (ix86_build_signbit_mask): Likewise.
7696 (ix86_match_ccmode): Likewise.
7697 (ix86_cc_modes_compatible): Likewise.
7698 (ix86_expand_branch): Likewise.
7699 (ix86_expand_sse_cmp): Likewise.
7700 (ix86_expand_sse_movcc): Likewise.
7701 (ix86_expand_int_sse_cmp): Likewise.
7702 (ix86_expand_vec_perm_vpermi2): Likewise.
7703 (ix86_expand_vec_perm): Likewise.
7704 (ix86_expand_sse_unpack): Likewise.
7705 (ix86_expand_int_addcc): Likewise.
7706 (ix86_split_to_parts): Likewise.
7707 (ix86_vectorize_builtin_gather): Likewise.
7708 (ix86_vectorize_builtin_scatter): Likewise.
7709 (avx_vpermilp_parallel): Likewise.
7710 (inline_memory_move_cost): Likewise.
7711 (ix86_tieable_integer_mode_p): Likewise.
7712 (x86_maybe_negate_const_int): Likewise.
7713 (ix86_expand_vector_init_duplicate): Likewise.
7714 (ix86_expand_vector_init_one_nonzero): Likewise.
7715 (ix86_expand_vector_init_one_var): Likewise.
7716 (ix86_expand_vector_init_concat): Likewise.
7717 (ix86_expand_vector_init_interleave): Likewise.
7718 (ix86_expand_vector_init_general): Likewise.
7719 (ix86_expand_vector_set): Likewise.
7720 (ix86_expand_vector_extract): Likewise.
7721 (emit_reduc_half): Likewise.
7722 (ix86_emit_i387_round): Likewise.
7723 (ix86_mangle_type): Likewise.
7724 (ix86_expand_round_sse4): Likewise.
7725 (expand_vec_perm_blend): Likewise.
7726 (canonicalize_vector_int_perm): Likewise.
7727 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
7728 (expand_vec_perm_1): Likewise.
7729 (expand_vec_perm_interleave3): Likewise.
7730 (expand_vec_perm_even_odd_pack): Likewise.
7731 (expand_vec_perm_even_odd_1): Likewise.
7732 (expand_vec_perm_broadcast_1): Likewise.
7733 (ix86_vectorize_vec_perm_const_ok): Likewise.
7734 (ix86_expand_vecop_qihi): Likewise.
7735 (ix86_expand_mul_widen_hilo): Likewise.
7736 (ix86_expand_sse2_abs): Likewise.
7737 (ix86_expand_pextr): Likewise.
7738 (ix86_expand_pinsr): Likewise.
7739 (ix86_preferred_simd_mode): Likewise.
7740 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
7741 * config/i386/sse.md (*andnot<mode>3): Likewise.
7742 (<mask_codefor><code><mode>3<mask_name>): Likewise.
7743 (*<code><mode>3): Likewise.
7744 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
7745 (ia64_expand_atomic_op): Likewise.
7746 (ia64_arg_type): Likewise.
7747 (ia64_mode_to_int): Likewise.
7748 (ia64_scalar_mode_supported_p): Likewise.
7749 (ia64_vector_mode_supported_p): Likewise.
7750 (expand_vec_perm_broadcast): Likewise.
7751 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
7752 (iq2000_function_arg_advance): Likewise.
7753 (iq2000_function_arg): Likewise.
7754 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
7755 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
7756 (m68k_libcall_value): Likewise.
7757 (m68k_function_value): Likewise.
7758 (sched_attr_op_type): Likewise.
7759 * config/mcore/mcore.c (mcore_output_move): Likewise.
7760 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
7762 (microblaze_function_arg): Likewise.
7763 * config/mips/mips.c (mips16_build_call_stub): Likewise.
7764 (mips_print_operand): Likewise.
7765 (mips_mode_ok_for_mov_fmt_p): Likewise.
7766 (mips_vector_mode_supported_p): Likewise.
7767 (mips_preferred_simd_mode): Likewise.
7768 (mips_expand_vpc_loongson_even_odd): Likewise.
7769 (mips_expand_vec_unpack): Likewise.
7770 (mips_expand_vi_broadcast): Likewise.
7771 (mips_expand_vector_init): Likewise.
7772 (mips_expand_vec_reduc): Likewise.
7773 (mips_expand_msa_cmp): Likewise.
7774 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
7775 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
7776 (cc_flags_for_mode): Likewise.
7777 * config/msp430/msp430.c (msp430_print_operand): Likewise.
7778 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
7779 (nds32_output_casesi_pc_relative): Likewise.
7780 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7781 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
7782 (nvptx_gen_unpack): Likewise.
7783 (nvptx_gen_pack): Likewise.
7784 (nvptx_gen_shuffle): Likewise.
7785 (nvptx_gen_wcast): Likewise.
7786 (nvptx_preferred_simd_mode): Likewise.
7787 * config/pa/pa.c (pa_secondary_reload): Likewise.
7788 * config/pa/predicates.md (base14_operand): Likewise.
7789 * config/powerpcspe/powerpcspe-c.c
7790 (altivec_resolve_overloaded_builtin): Likewise.
7791 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
7793 (rs6000_preferred_simd_mode): Likewise.
7794 (output_vec_const_move): Likewise.
7795 (rs6000_expand_vector_extract): Likewise.
7796 (rs6000_split_vec_extract_var): Likewise.
7797 (reg_offset_addressing_ok_p): Likewise.
7798 (rs6000_legitimate_offset_address_p): Likewise.
7799 (rs6000_legitimize_address): Likewise.
7800 (rs6000_emit_set_const): Likewise.
7801 (rs6000_const_vec): Likewise.
7802 (rs6000_emit_move): Likewise.
7803 (spe_build_register_parallel): Likewise.
7804 (rs6000_darwin64_record_arg_recurse): Likewise.
7805 (swap_selector_for_mode): Likewise.
7806 (spe_init_builtins): Likewise.
7807 (paired_init_builtins): Likewise.
7808 (altivec_init_builtins): Likewise.
7809 (do_load_for_compare): Likewise.
7810 (rs6000_generate_compare): Likewise.
7811 (rs6000_expand_float128_convert): Likewise.
7812 (emit_load_locked): Likewise.
7813 (emit_store_conditional): Likewise.
7814 (rs6000_output_function_epilogue): Likewise.
7815 (rs6000_handle_altivec_attribute): Likewise.
7816 (rs6000_function_value): Likewise.
7817 (emit_fusion_gpr_load): Likewise.
7818 (emit_fusion_p9_load): Likewise.
7819 (emit_fusion_p9_store): Likewise.
7820 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
7821 (fusion_gpr_mem_load): Likewise.
7822 (fusion_addis_mem_combo_load): Likewise.
7823 (fusion_addis_mem_combo_store): Likewise.
7824 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
7825 (fusion_gpr_mem_load): Likewise.
7826 (fusion_addis_mem_combo_load): Likewise.
7827 (fusion_addis_mem_combo_store): Likewise.
7828 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7830 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
7831 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
7832 (rs6000_preferred_simd_mode): Likewise.
7833 (output_vec_const_move): Likewise.
7834 (rs6000_expand_vector_extract): Likewise.
7835 (rs6000_split_vec_extract_var): Likewise.
7836 (reg_offset_addressing_ok_p): Likewise.
7837 (rs6000_legitimate_offset_address_p): Likewise.
7838 (rs6000_legitimize_address): Likewise.
7839 (rs6000_emit_set_const): Likewise.
7840 (rs6000_const_vec): Likewise.
7841 (rs6000_emit_move): Likewise.
7842 (rs6000_darwin64_record_arg_recurse): Likewise.
7843 (swap_selector_for_mode): Likewise.
7844 (paired_init_builtins): Likewise.
7845 (altivec_init_builtins): Likewise.
7846 (rs6000_expand_float128_convert): Likewise.
7847 (emit_load_locked): Likewise.
7848 (emit_store_conditional): Likewise.
7849 (rs6000_output_function_epilogue): Likewise.
7850 (rs6000_handle_altivec_attribute): Likewise.
7851 (rs6000_function_value): Likewise.
7852 (emit_fusion_gpr_load): Likewise.
7853 (emit_fusion_p9_load): Likewise.
7854 (emit_fusion_p9_store): Likewise.
7855 * config/rx/rx.c (rx_gen_move_template): Likewise.
7856 (flags_from_mode): Likewise.
7857 * config/s390/predicates.md (s390_alc_comparison): Likewise.
7858 (s390_slb_comparison): Likewise.
7859 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
7860 (s390_vector_mode_supported_p): Likewise.
7861 (s390_cc_modes_compatible): Likewise.
7862 (s390_match_ccmode_set): Likewise.
7863 (s390_canonicalize_comparison): Likewise.
7864 (s390_emit_compare_and_swap): Likewise.
7865 (s390_branch_condition_mask): Likewise.
7866 (s390_rtx_costs): Likewise.
7867 (s390_secondary_reload): Likewise.
7868 (__SECONDARY_RELOAD_CASE): Likewise.
7869 (s390_expand_cs): Likewise.
7870 (s390_preferred_simd_mode): Likewise.
7871 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
7872 * config/sh/sh.c (sh_print_operand): Likewise.
7873 (dump_table): Likewise.
7874 (sh_secondary_reload): Likewise.
7875 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7876 * config/sh/sh.md (casesi_worker_1): Likewise.
7877 (casesi_worker_2): Likewise.
7878 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
7879 (fcc_comparison_operator): Likewise.
7880 * config/sparc/sparc.c (sparc_expand_move): Likewise.
7881 (emit_soft_tfmode_cvt): Likewise.
7882 (sparc_preferred_simd_mode): Likewise.
7883 (output_cbranch): Likewise.
7884 (sparc_print_operand): Likewise.
7885 (sparc_expand_vec_perm_bmask): Likewise.
7886 (vector_init_bshuffle): Likewise.
7887 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
7888 (spu_vector_mode_supported_p): Likewise.
7889 (spu_expand_insv): Likewise.
7890 (spu_emit_branch_or_set): Likewise.
7891 (spu_handle_vector_attribute): Likewise.
7892 (spu_builtin_splats): Likewise.
7893 (spu_builtin_extract): Likewise.
7894 (spu_builtin_promote): Likewise.
7895 (spu_expand_sign_extend): Likewise.
7896 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
7897 (tilegx_simd_int): Likewise.
7898 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
7899 (tilepro_simd_int): Likewise.
7900 * config/v850/v850.c (const_double_split): Likewise.
7901 (v850_print_operand): Likewise.
7902 (ep_memory_offset): Likewise.
7903 * config/vax/vax.c (vax_rtx_costs): Likewise.
7904 (vax_output_int_move): Likewise.
7905 (vax_output_int_add): Likewise.
7906 (vax_output_int_subtract): Likewise.
7907 * config/visium/predicates.md (visium_branch_operator): Likewise.
7908 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
7909 (visium_print_operand_address): Likewise.
7910 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7911 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
7912 (xtensa_expand_conditional_branch): Likewise.
7913 (xtensa_copy_incoming_a7): Likewise.
7914 (xtensa_output_literal): Likewise.
7915 * dfp.c (decimal_real_maxval): Likewise.
7916 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
7918 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7919 Alan Hayward <alan.hayward@arm.com>
7920 David Sherwood <david.sherwood@arm.com>
7922 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
7923 (mode_nunits_inline): Likewise.
7924 (mode_inner_inline): Likewise.
7925 (mode_unit_size_inline): Likewise.
7926 (mode_unit_precision_inline): Likewise.
7927 (emit_insn_modes_h): Likewise. Also emit a #define of the
7929 (emit_mode_wider): Add an E_ prefix to mode names.
7930 (emit_mode_complex): Likewise.
7931 (emit_mode_inner): Likewise.
7932 (emit_mode_adjustments): Likewise.
7933 (emit_mode_int_n): Likewise.
7934 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
7935 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
7936 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
7937 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
7938 (CRC32_BUILTIN, ENTRY): Likewise.
7939 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
7940 (aarch64_pop_regs): Likewise.
7941 (aarch64_process_components): Likewise.
7942 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
7943 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
7944 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
7945 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
7946 * config/arm/arm.c (arm_init_libfuncs): Likewise.
7947 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
7949 * config/i386/i386-builtin.def (pcmpestr): Likewise.
7950 (pcmpistr): Likewise.
7951 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
7952 * config/mmix/mmix.c (mmix_output_condition): Likewise.
7953 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
7955 * config/rl78/rl78.c (mduc_regs): Likewise.
7956 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
7957 (htm_expand_builtin): Likewise.
7958 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
7959 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
7960 * config/xtensa/xtensa.c (print_operand): Likewise.
7961 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
7962 (NUM_MODE_VECTOR_INT): Likewise.
7963 * genoutput.c (null_operand): Likewise.
7964 (output_operand_data): Likewise.
7965 * genrecog.c (print_parameter_value): Likewise.
7966 * lra.c (debug_operand_data): Likewise.
7968 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7970 * dwarf2out.c (scompare_loc_descriptor_wide)
7971 (scompare_loc_descriptor_narrow): New functions, split out from...
7972 (scompare_loc_descriptor): ...here.
7973 * expmed.c (emit_store_flag_int): New function, split out from...
7974 (emit_store_flag): ...here.
7976 2017-08-30 Richard Biener <rguenther@suse.de>
7978 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
7979 (dwarf2out_early_finish): Move setting of AT_pubnames from
7980 early debug output to early finish.
7982 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
7984 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
7985 and -mdata-region to the assembler.
7987 2017-08-30 Richard Biener <rguenther@suse.de>
7989 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
7991 (gen_subprogram_die): Add DW_AT_object_pointer only early.
7992 (dwarf2out_early_global_decl): Only generate a DIE for the
7993 abstract origin if it doesn't already exist or is a declaration DIE.
7994 (resolve_addr): Do not add the linkage name twice when
7995 generating a stub DIE for the DW_TAG_GNU_call_site target.
7997 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7999 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8000 Use machine_mode rather than int for arg1_mode.
8002 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
8005 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
8006 that the second argument of the built-in functions to unpack
8007 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
8008 switch statement instead a lot of if statements.
8009 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
8010 Allow 64-bit values to be in Altivec registers as well as
8011 traditional floating point registers.
8012 (pack<mode>, FMOVE128_VSX iterator): Likewise.
8014 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
8016 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
8017 MAX_REGS_PER_ADDRESS == 1.
8019 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
8021 * config/i386/i386.opt (flag_fentry): Do not init to -1.
8022 * config/i386/i386.c (ix86_option_override_internal): Simplify
8023 setting of opts->x_flag_entry.
8025 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8026 Jakub Jelinek <jakub@redhat.com>
8027 Richard Biener <rguenther@suse.de>
8029 PR tree-optimization/81503
8030 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
8031 folded constant fits in the target type; reorder tests for clarity.
8033 2017-08-29 Martin Liska <mliska@suse.cz>
8035 * passes.def: Include pass_lower_switch.
8036 * stmt.c (dump_case_nodes): Remove and move to
8037 tree-switch-conversion.
8038 (case_values_threshold): Likewise.
8039 (expand_switch_as_decision_tree_p): Likewise.
8040 (emit_case_decision_tree): Likewise.
8041 (expand_case): Likewise.
8042 (balance_case_nodes): Likewise.
8043 (node_has_low_bound): Likewise.
8044 (node_has_high_bound): Likewise.
8045 (node_is_bounded): Likewise.
8046 (emit_case_nodes): Likewise.
8047 (struct simple_case_node): New struct.
8048 (add_case_node): Remove.
8049 (emit_case_dispatch_table): Use vector instead of case_list.
8050 (reset_out_edges_aux): Remove.
8051 (compute_cases_per_edge): Likewise.
8052 (expand_case): Build list of simple_case_node.
8053 (expand_sjlj_dispatch_table): Use it.
8054 * tree-switch-conversion.c (struct case_node): Moved from
8055 stmt.c and adjusted.
8056 (emit_case_nodes): Likewise.
8057 (node_has_low_bound): Likewise.
8058 (node_has_high_bound): Likewise.
8059 (node_is_bounded): Likewise.
8060 (case_values_threshold): Likewise.
8061 (reset_out_edges_aux): Likewise.
8062 (compute_cases_per_edge): Likewise.
8063 (add_case_node): Likewise.
8064 (dump_case_nodes): Likewise.
8065 (balance_case_nodes): Likewise.
8066 (expand_switch_as_decision_tree_p): Likewise.
8067 (emit_jump): Likewise.
8068 (emit_case_decision_tree): Likewise.
8069 (try_switch_expansion): Likewise.
8070 (do_jump_if_equal): Likewise.
8071 (emit_cmp_and_jump_insns): Likewise.
8072 (fix_phi_operands_for_edge): New function.
8073 (record_phi_operand_mapping): Likewise.
8074 (class pass_lower_switch): New pass.
8075 (pass_lower_switch::execute): New function.
8076 (make_pass_lower_switch): Likewise.
8077 (conditional_probability):
8078 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
8079 * tree-pass.h: Add make_pass_lower_switch.
8081 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
8084 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
8087 2017-08-29 Richard Biener <rguenther@suse.de>
8089 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
8090 we do not add a DW_AT_inline attribute twice.
8091 (gen_subprogram_die): Remove code setting DW_AT_inline on
8092 DECL_ABSTRACT_P nodes.
8094 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
8096 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
8097 calls to internal functions.
8098 (gimplify_modify_expr): Likewise.
8099 * tree-call-cdce.c (use_internal_fn): Likewise.
8100 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
8101 (convert_to_divmod): Set the nothrow flag.
8102 * tree-if-conv.c (predicate_mem_writes): Likewise.
8103 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
8104 (vectorizable_call): Likewise.
8105 (vectorizable_store): Likewise.
8106 (vectorizable_load): Likewise.
8107 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
8108 (vect_recog_mask_conversion_pattern): Likewise.
8110 2017-08-29 Martin Liska <mliska@suse.cz>
8113 * gcc.c (driver_handle_option): Add new argument.
8114 * opts-common.c (handle_option): Pass
8115 target_option_override_hook.
8116 * opts-global.c (lang_handle_option): Add new option.
8117 (set_default_handlers): Add new argument.
8118 (decode_options): Likewise.
8119 * opts.c (target_handle_option): Likewise.
8120 (common_handle_option): Call target_option_override_hook.
8121 * opts.h (struct cl_option_handler_func): Add hook for
8122 target option override.
8123 (struct cl_option_handlers): Likewise.
8124 (set_default_handlers): Add new argument.
8125 (decode_options): Likewise.
8126 (common_handle_option): Likewise.
8127 (target_handle_option): Likewise.
8128 * toplev.c (toplev::main): Pass targetm.target_option.override
8131 2017-08-29 Richard Biener <rguenther@suse.de>
8132 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
8134 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
8135 life to the active subtree.
8137 2017-08-28 Jeff Law <law@redhat.com>
8139 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
8140 derive_equivalences.
8141 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
8143 (edge_info::derive_equivalences): New private member function
8145 * tree-ssa-dom.c (class edge_info): Changed from a struct
8146 to a class. Add ctor/dtor, methods and data members.
8147 (edge_info::edge_info): Renamed from allocate_edge_info.
8148 Initialize additional members.
8149 (edge_info::~edge_info): New.
8150 (free_dom_edge_info): Delete the edge info.
8151 (record_edge_info): Use new class & associated member functions.
8152 Tighten forms for testing for edge equivalences.
8153 (record_temporary_equivalences): Iterate over the simple
8154 equivalences rather than assuming there's only one per edge.
8155 (cprop_into_successor_phis): Iterate over the simple
8156 equivalences rather than assuming there's only one per edge.
8157 (optimize_stmt): Use operand_equal_p rather than pointer
8158 equality for mini-DSE code.
8160 2017-08-28 Nathan Sidwell <nathan@acm.org>
8162 * gcc.c (execute): Fold SIGPIPE handling into switch
8163 statement. Adjust internal error message.
8165 2017-08-28 Richard Biener <rguenther@suse.de>
8168 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
8169 Do nothing for removed DIEs.
8171 2017-08-28 Richard Biener <rguenther@suse.de>
8173 PR tree-optimization/81977
8174 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
8177 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
8180 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
8181 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
8182 using targetm.gen_mem_thread_fence.
8184 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
8187 * config/i386/i386.md (*<btsc><mode>): Change operand 2
8188 predicate to register_operand. Reorder operands.
8189 (*btr<mode>): Ditto.
8190 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
8191 (*btr<mode>_mask): Ditto.
8193 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
8195 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
8196 * config/rs6000/xmmintrin.h: New file.
8197 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
8199 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8202 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
8203 parameter and_insn and return it.
8204 (recombine_lvx_pattern): Insert a copy to ensure availability of
8205 the base register of the copied masking operation at the point of
8206 the instruction replacement.
8207 (recombine_stvx_pattern): Likewise.
8209 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
8211 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
8212 undocumented switches.
8213 (-mpower9-dform-vector): Likewise.
8214 (-mpower9-dform): Likewise.
8215 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
8216 comments to delete references to -mpower9-dform* switches.
8217 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
8218 Delete reference to -mpower9-dform* switches, test for
8219 -mpower9-vector instead.
8220 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
8221 (OTHER_P9_VECTOR_MASKS): Likewise.
8222 (POWERPC_MASKS): Likewise.
8223 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
8224 tests against -mpower9-dform* to -mpower9-vector. Delete code
8225 that checked for -mpower9-dform* consistancy with other options.
8226 Add test for -mpower9-misc to enable other power9 switches.
8227 (rs6000_init_hard_regno_mode_ok): Likewise.
8228 (rs6000_option_override_internal): Likewise.
8229 (rs6000_emit_prologue): Likewise.
8230 (rs6000_emit_epilogue): Likewise.
8231 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
8232 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
8233 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
8235 (emit_fusion_p9_store): Likewise.
8236 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
8237 resetting these macros if the assembler does not support ISA 3.0
8239 (TARGET_P9_DFORM_VECTOR): Likewise.
8240 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
8241 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
8243 2017-08-25 Alan Modra <amodra@gmail.com>
8245 PR rtl-optimization/81747
8246 * cse.c (cse_extended_basic_block): Don't attempt to record
8247 equivalences for degenerate conditional branches that branch
8248 to their fall-through.
8250 2017-08-24 Martin Sebor <msebor@redhat.com>
8253 * gimple-fold.c (size_must_be_zero_p): New function.
8254 (gimple_fold_builtin_memory_op): Call it.
8256 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
8258 * config/rs6000/mm_malloc.h: New file.
8260 2017-08-24 Bin Cheng <bin.cheng@arm.com>
8262 PR tree-optimization/81913
8263 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
8264 analysis when either IVs in condition can wrap.
8266 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
8268 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
8269 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
8271 2017-08-24 Richard Biener <rguenther@suse.de>
8274 * targhooks.c (default_target_can_inline_p): Properly
8275 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
8276 is present and always compare.
8277 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
8278 infer -mfpmath=sse from TARGET_SSE_P.
8279 (ix86_can_inline_p): Properly use target_option_default_node when
8280 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
8282 2017-08-24 Richard Biener <rguenther@suse.de>
8285 * dwarf2out.c (output_die): Handle flag_generate_offload like
8287 (output_comp_unit): Likewise.
8288 (gen_array_type_die): Likewise.
8289 (dwarf2out_early_finish): Likewise.
8290 (note_variable_value_in_expr): Likewise.
8291 (dwarf2out_finish): Likewise. Adjust assert.
8292 * cgraphunit.c (symbol_table::compile): Move setting of
8293 flag_generate_offload earlier ...
8294 (symbol_table::finalize_compilation_unit): ... here, before
8295 early debug finalization.
8297 2017-08-24 Richard Biener <rguenther@suse.de>
8299 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
8300 and ipa-fnsummary.h.
8301 (ix86_can_inline_p): When ix86_fpmath flags do not match
8302 check whether the callee uses FP math at all.
8304 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
8307 * tree-ssanames.c (get_nonzero_bits): Use element_precision
8308 instead of TYPE_PRECISION.
8310 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
8311 Alan Hayward <alan.hayward@arm.com>
8312 David Sherwood <david.sherwood@arm.com>
8314 * combine.c (make_extraction): Use subreg_offset_from_lsb.
8316 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
8318 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
8320 (ix86_frame::stack_realign_allocate): New field.
8321 (struct machine_frame_state): Modify comments.
8322 (machine_frame_state::sp_realigned_fp_end): New field.
8323 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
8325 (sp_valid_at): Add assertion to assure no attempt to access invalid
8326 offset of a realigned stack.
8327 (fp_valid_at): Likewise.
8328 (choose_baseaddr): Modify comments.
8329 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
8330 ix86_expand_prologue.
8331 (ix86_expand_prologue): Modify stack realignment and allocation.
8332 (ix86_expand_epilogue): Modify comments.
8333 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
8334 avx2_runtime, avx512f, and avx512f_runtime.
8336 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
8338 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
8339 (mstackrealign): Do not init to -1.
8340 * config/i386/i386.c (ix86_option_override_internal):
8341 Check opts_set, not opts when setting default value of
8342 opts->x_ix86_force_align_arg_pointer.
8344 2017-08-23 Richard Biener <rguenther@suse.de>
8346 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
8347 lang_hooks.decl_printable_name.
8348 * print-rtl-function.c (print_rtx_function): Likewise.
8349 * tree-pretty-print.c (dump_function_header): Likewise.
8351 2017-08-23 Richard Biener <rguenther@suse.de>
8354 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
8355 -g0 at compile-time.
8357 2017-08-23 Tamar Christina <tamar.christina@arm.com>
8360 * doc/sourcebuild.texi (Other hardware attributes):
8363 2017-08-23 Tamar Christina <tamar.christina@arm.com>
8366 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
8367 Fix single-use check.
8369 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
8371 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
8373 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
8375 * doc/install.texi: Modify to add more details on running selected
8378 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
8380 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
8381 is combined with -mabi=ms.
8382 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
8385 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8387 PR tree-optimization/81488
8388 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
8389 and cached_basis fields.
8390 (MAX_SPREAD): New constant.
8391 (alloc_cand_and_find_basis): Initialize new fields.
8392 (clear_visited): New function.
8393 (create_phi_basis_1): Rename from create_phi_basis, set visited
8394 and cached_basis fields.
8395 (create_phi_basis): New wrapper function.
8396 (phi_add_costs_1): Rename from phi_add_costs, add spread
8397 parameter, set visited field, short-circuit when limits reached.
8398 (phi_add_costs): New wrapper function.
8399 (record_phi_increments_1): Rename from record_phi_increments, set
8401 (record_phi_increments): New wrapper function.
8402 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
8403 (phi_incr_cost): New wrapper function.
8404 (all_phi_incrs_profitable_1): Rename from
8405 all_phi_incrs_profitable, set visited field.
8406 (all_phi_incrs_profitable): New wrapper function.
8408 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
8409 Alan Hayward <alan.hayward@arm.com>
8410 David Sherwood <david.sherwood@arm.com>
8412 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
8413 that takes the outer and inner modes.
8414 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
8415 comparison as the canonical test for a paradoxical subreg.
8416 * combine.c (simplify_set): Use paradoxical_subreg_p.
8417 (make_extraction): Likewise.
8418 (force_to_mode): Likewise.
8419 (rtx_equal_for_field_assignment_p): Likewise.
8420 (gen_lowpart_for_combine): Likewise.
8421 (simplify_comparison): Likewise.
8422 * cse.c (equiv_constant): Likewise.
8423 * expmed.c (store_bit_field_1): Likewise.
8424 * final.c (alter_subreg): Likewise.
8425 * fwprop.c (propagate_rtx): Likewise.
8426 (forward_propagate_subreg): Likewise.
8427 * ira-conflicts.c (ira_build_conflicts): Likewise.
8428 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
8429 * lra-constraints.c (curr_insn_transform): Likewise.
8430 (split_reg): Likewise.
8431 * lra-eliminations.c (move_plus_up): Likewise.
8432 (lra_eliminate_regs_1): Likewise.
8433 * recog.c (general_operand): Likewise.
8434 * ree.c (combine_reaching_defs): Likewise.
8435 * reload.c (push_reload): Likewise.
8436 (find_reloads): Likewise.
8437 * reload1.c (elimination_effects): Likewise.
8438 (compute_reload_subreg_offset): Likewise.
8439 (choose_reload_regs): Likewise.
8440 * rtlanal.c (subreg_lsb_1): Likewise.
8441 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8442 (simplify_subreg): Likewise.
8443 * var-tracking.c (track_loc_p): Likewise.
8444 * emit-rtl.c (byte_lowpart_offset): Likewise.
8445 (paradoxical_subreg_p): Delete out-of-line definition.
8447 2017-08-22 Jeff Law <law@redhat.com>
8449 PR tree-optimization/81741
8450 PR tree-optimization/71947
8451 * tree-ssa-dom.c: Include tree-inline.h.
8452 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
8453 equivalences if one is more expensive to compute than the other.
8454 * tree-ssa-scopedtables.h (class const_or_copies): Make
8455 record_const_or_copy_raw method private.
8456 (class avail_exprs_stack): New method simplify_binary_operation.
8457 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
8458 avail_exprs_stack::simplify_binary_operation as needed.
8459 (avail_exprs_stack::simplify_binary_operation): New function.
8461 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
8463 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
8464 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
8465 (DOT_SYMBOLS): Likewise.
8466 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
8467 (RELOCATABLE_NEEDS_FIXUP): Likewise.
8468 (RS6000_ABI_NAME): Likewise.
8469 (TARGET_CMODEL): Likewise.
8470 (TOC_SECTION_ASM_OP): Likewise.
8471 (SET_CMODEL): New macro.
8472 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
8474 2017-08-22 Richard Biener <rguenther@suse.de>
8476 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
8477 to allow for free-lang-data replacements similar to verify_type_variant.
8479 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
8480 Alan Hayward <alan.hayward@arm.com>
8481 David Sherwood <david.sherwood@arm.com>
8483 * config/aarch64/aarch64.md (casesi): Use DImode rather than
8484 VOIDmode for the LABEL_REF.
8486 2017-08-22 Richard Biener <rguenther@suse.de>
8488 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
8490 2017-08-22 Richard Biener <rguenther@suse.de>
8492 * common.opt (feliminate-dwarf2-dups): Ignore.
8493 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
8494 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
8495 same_die_p_wrap, compute_section_prefix,
8496 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
8497 (comdat_symbol_id, comdat_symbol_number): Likewise.
8498 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
8500 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
8501 (output_die): Mark unreachable path unreachable.
8502 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
8503 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
8504 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
8505 (dwarf2out_early_finish): Likewise.
8507 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
8509 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
8511 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
8514 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
8515 not VAR_P. Filter attribute warnings with OPT_Wattributes.
8516 (avr_attribute_table) <io, io_low, address>: Initialize
8517 .decl_required with true.
8519 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
8521 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
8522 undocumented debugging options.
8523 (-mvsx-scalar-double): Likewise.
8524 (-mallow-df-permute): Likewise.
8525 (-mvectorize-builtins): Likewise.
8526 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8527 (rs6000_builtin_vectorized_function): Likewise.
8528 (rs6000_builtin_md_vectorized_function): Likewise.
8529 (rs6000_opt_vars): Likewise.
8531 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
8534 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
8535 (*btrq_imm): Rename from *btrq.
8536 (*btcq_imm): Rename from *btcq.
8537 (btsc): New code attribute.
8538 (*<btsc><mode>): New insn pattern.
8539 (*btr<mode>): Ditto.
8540 (*<btsc><mode>_mask): New insn_and_split pattern.
8541 (*btr<mode>_mask): Ditto.
8543 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8544 Alan Hayward <alan.hayward@arm.com>
8545 David Sherwood <david.sherwood@arm.com>
8547 * function.c (pad_below): Simplify padding calculation.
8549 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8550 Alan Hayward <alan.hayward@arm.com>
8551 David Sherwood <david.sherwood@arm.com>
8553 * target.def (function_prologue): Remove frame size argument.
8554 (function_epilogue): Likewise.
8555 * doc/tm.texi: Regenerate.
8556 * final.c (final_start_function): Update call to function_prologue.
8557 (final_end_function): Update call to function_epilogue.
8558 (default_function_pro_epilogue): Remove frame size argument.
8559 * output.h (default_function_pro_epilogue): Likewise.
8560 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
8561 (arm_output_function_prologue): Likewise.
8562 * config/frv/frv.c (frv_function_prologue): Likewise.
8563 (frv_function_epilogue): Likewise.
8564 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
8565 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
8566 (ia64_output_function_epilogue): Likewise.
8567 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
8568 (m32r_output_function_epilogue): Likewise.
8569 * config/microblaze/microblaze.c (microblaze_function_prologue)
8570 (microblaze_function_epilogue): Likewise.
8571 * config/mips/mips.c (mips_output_function_prologue): Likewise.
8572 (mips_output_function_epilogue): Likewise.
8573 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
8574 (mmix_target_asm_function_epilogue): Likewise.
8575 * config/msp430/msp430.c (msp430_start_function): Likewise.
8576 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
8577 (nds32_asm_function_epilogue): Likewise.
8578 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
8579 * config/pa/pa.c (pa_output_function_prologue): Likewise.
8580 (pa_output_function_epilogue): Likewise.
8581 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
8582 (rs6000_output_function_epilogue): Likewise.
8583 * config/rl78/rl78.c (rl78_start_function): Likewise.
8584 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
8585 (rs6000_output_function_epilogue): Likewise.
8586 * config/rx/rx.c (rx_output_function_prologue): Likewise.
8587 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
8588 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
8589 (sparc_asm_function_epilogue): Likewise.
8591 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8593 * tree.h (type_has_mode_precision_p): New function.
8594 * convert.c (convert_to_integer_1): Use it.
8595 * expr.c (expand_expr_real_2): Likewise.
8596 (expand_expr_real_1): Likewise.
8597 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
8598 * match.pd: Likewise.
8599 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
8600 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
8601 * tree-tailcall.c (process_assignment): Likewise.
8602 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8603 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
8604 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
8605 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8606 (vectorizable_assignment): Likewise.
8607 (vectorizable_shift): Likewise.
8608 (vectorizable_operation): Likewise.
8609 * tree-vrp.c (register_edge_assert_for_2): Likewise.
8611 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
8613 * match.pd: Add pow (C, x) simplification.
8615 2017-08-21 Richard Biener <rguenther@suse.de>
8617 PR tree-optimization/81900
8618 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
8619 for blocks with abnormal predecessors.
8620 (compute_antic): Do not set visited flag prematurely.
8622 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
8625 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
8627 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8629 * stor-layout.h (vector_type_mode): Move to...
8630 * tree.h (vector_type_mode): ...here.
8631 * stor-layout.c (vector_type_mode): Move to...
8632 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
8634 2017-08-21 Richard Biener <rguenther@suse.de>
8636 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
8637 register_external_die hooks.
8638 (debug_false_tree_charstarstar_uhwistar): Declare.
8639 (debug_nothing_tree_charstar_uhwi): Likewise.
8640 * debug.c (do_nothing_debug_hooks): Adjust.
8641 (debug_false_tree_charstarstar_uhwistar): New do nothing.
8642 (debug_nothing_tree_charstar_uhwi): Likewise.
8643 * dbxout.c (dbx_debug_hooks): Adjust.
8644 (xcoff_debug_hooks): Likewise.
8645 * sdbout.c (sdb_debug_hooks): Likewise.
8646 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
8647 * dwarf2out.c (macinfo_label_base): New global.
8648 (dwarf2out_register_external_die): New function for the
8649 register_external_die hook.
8650 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
8651 (dwarf2_debug_hooks): Use them.
8652 (dwarf2_lineno_debug_hooks): Adjust.
8653 (struct die_struct): Add with_offset flag.
8654 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
8655 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
8656 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
8657 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
8658 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
8659 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
8660 defining section names for the early LTO debug variants.
8661 (reset_indirect_string): New helper.
8662 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
8663 (print_dw_val): Add support for offsetted symbol references.
8664 (get_ultimate_context): Split out from is_cxx.
8665 (is_cxx): Use get_ultimate_context.
8666 (is_fortran): Add decl overload.
8667 (compute_comp_unit_symbol): Split out worker from
8668 compute_section_prefix.
8669 (compute_section_prefix): Call compute_comp_unit_symbol and
8670 set comdat_type_p here.
8671 (output_die): Skip DIE symbol output for the LTO added one.
8672 Handle DIE symbol references with offset.
8673 (output_comp_unit): Guard section name mangling properly.
8674 For LTO debug sections emit a symbol at the section beginning
8675 which we use to refer to its DIEs.
8676 (add_abstract_origin_attribute): For DIEs registered via
8677 dwarf2out_register_external_die directly refer to the early
8678 DIE rather than indirectly through the shadow one we created.
8679 Remove obsolete call to dwarf2out_abstract_function for
8680 non-function/block origins.
8681 (gen_array_type_die): When generating early LTO debug do
8682 not emit DW_AT_string_length.
8683 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
8684 late when in LTO. As suggested place a gcc_unreachable for
8685 the DECL_ABSTRACT_P case.
8686 (gen_subprogram_die): Avoid another specification DIE
8687 for early built declarations/definitions for the late LTO case.
8688 (gen_variable_die): Add type references for late duplicated VLA dies
8690 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
8691 we have the abstract instance already.
8692 (process_scope_var): Adjust decl DIE contexts in LTO which
8693 first puts them in limbo.
8694 (gen_decl_die): Do not generate type DIEs late apart from
8695 types for VLAs or for decls we do not yet have a DIE. Do not
8696 call dwarf2out_abstract_function late.
8697 (dwarf2out_early_global_decl): Make sure to create DIEs
8698 for abstract instances of a decl first.
8699 (dwarf2out_late_global_decl): Adjust comment.
8700 (output_macinfo_op): With multiple macro sections use
8701 macinfo_label_base to distinguish labels.
8702 (output_macinfo): Likewise. Update macinfo_label_base.
8703 Pass in the line info label.
8704 (note_variable_value_in_expr): When generating LTO resolve
8705 all variable values here by generating DIEs as needed.
8706 (init_sections_and_labels): Add early LTO debug flag parameter
8707 and generate different sections and names if set. Add generation
8708 counter for the labels so we can have multiple of them.
8709 (reset_dies): Helper to allow DIEs to be output multiple times.
8710 (dwarf2out_finish): When outputting DIEs to the fat part of an
8711 LTO object first reset DIEs.
8712 (dwarf2out_early_finish): Output early DIEs when generating LTO.
8713 (modified_type_die): Check for decl_ultimate_origin being self
8715 (gen_type_die_with_usage): Likewise.
8716 (gen_typedef_die): Allow decl_ultimate_origin being self.
8717 (set_decl_abstract_flags): Remove.
8718 (set_block_abstract_flags): Likewise.
8719 (dwarf2out_abstract_function): Treat the early generated DIEs
8720 as the abstract copy and only add DW_AT_inline and
8721 DW_AT_artificial here and call set_decl_origin_self.
8722 If the DIE has an abstract origin don't do anything.
8723 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
8724 if we have none yet (Go fails to build one, PR78628).
8725 (variably_modified_type_p): Prevent endless recursion for Ada
8726 cyclic pointer types.
8727 * lto-streamer-in.c: Include debug.h.
8728 (dref_queue): New global.
8729 (lto_read_tree_1): Stream in DIE references.
8730 (lto_input_tree): Register DIE references.
8731 (input_function): Stream DECL_DEBUG_ARGS.
8732 * lto-streamer-out.c: Include debug.h.
8733 (lto_write_tree_1): Output DIE references.
8734 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
8735 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
8736 (output_function): Stream DECL_DEBUG_ARGS.
8737 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
8738 Stream DECL_ABSTRACT_ORIGIN.
8739 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
8740 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
8741 DECL_CONTEXT for file-scope decls.
8742 * lto-streamer.h (struct dref_entry): Declare.
8743 (dref_queue): Likewise.
8744 * cfgexpand.c (pass_expand::execute): Do not call the
8745 outlining_inline_function hook here.
8746 * lto-wrapper.c (debug_obj): New global.
8747 (tool_cleanup): Unlink it if required.
8748 (debug_objcopy): New function.
8749 (run_gcc): Handle early debug sections in the IL files by
8750 extracting them to separate files, partially linkin them and
8751 feeding the result back as result to the linker.
8752 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
8753 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
8754 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
8755 sections into a separate segment.
8756 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
8758 (darwin_asm_dwarf_section): Likewise.
8759 (darwin_asm_output_dwarf_offset): Likewise.
8760 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
8762 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8763 Alan Hayward <alan.hayward@arm.com>
8764 David Sherwood <david.sherwood@arm.com>
8766 * read-md.h (md_reader::record_potential_iterator_use): Replace
8767 pointer argument with an rtx and an index.
8768 * read-rtl.c (iterator_group::apply_iterator): Likewise.
8769 (apply_mode_iterator): Likewise.
8770 (apply_code_iterator): Likewise.
8771 (apply_int_iterator): Likewise.
8772 (apply_subst_iterator): Likewise.
8773 (record_iterator_use): Likewise.
8774 (record_attribute_use): Likewise.
8775 (md_reader::record_potential_iterator_use): Likewise. Update calls
8776 to record_iterator_use and apply_iterator.
8777 (iterator_use): Replace ptr with x and index.
8778 (attribute_use): Likewise.
8779 (apply_attribute_uses): Update calls to apply_iterator.
8780 (apply_iterators): Likewise. Update initialization of iterator_use.
8781 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
8782 and record_potential_iterator_use.
8783 (rtx_reader::read_rtx_operand): Likewise.
8785 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8786 Alan Hayward <alan.hayward@arm.com>
8787 David Sherwood <david.sherwood@arm.com>
8789 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
8792 2017-08-21 Richard Biener <rguenther@suse.de>
8795 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
8796 at struct end conservatively when comparing common bases.
8798 2017-08-21 Richard Biener <rguenther@suse.de>
8800 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
8801 (mem_ref_in_stmt): Remove.
8802 (determine_max_movement): Use ref index to get at the reference.
8803 (invariantness_dom_walker::before_dom_children): Deal with
8804 lim data already initialized.
8805 (gather_mem_refs_stmt): Initialize lim data and record ref index.
8807 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
8809 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
8810 (TARGET_ISA_ROUND): Ditto.
8811 (TARGET_ROUND): Ditto.
8812 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
8813 * config/i386/i386.md: Ditto.
8814 * config/i386/sse.md: Ditto.
8815 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
8816 with OPTION_MASK_ISA_SSE4_1.
8818 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
8821 * doc/extend.texi (x86 Built-in Functions): Correct the name of
8822 __builtin_ia32_lzcnt_u16.
8824 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
8827 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
8828 (rs6000_set_current_function): Rewrite function to use it.
8830 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
8833 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
8834 and TYPE_WARN_IF_NOT_ALIGN.
8835 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
8836 (handle_warn_if_not_align): New.
8837 (place_union_field): Call handle_warn_if_not_align.
8838 (place_field): Call handle_warn_if_not_align.
8839 Copy TYPE_WARN_IF_NOT_ALIGN.
8840 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
8841 (layout_type): Likewise.
8842 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
8844 (tree_decl_common): Add warn_if_not_align.
8845 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
8846 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
8847 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
8848 (DECL_WARN_IF_NOT_ALIGN): Likewise.
8849 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
8850 * doc/extend.texi: Document warn_if_not_aligned attribute.
8851 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
8853 2017-08-17 Martin Liska <mliska@suse.cz>
8856 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
8857 (get_data_dependence): Use it as pointer type.
8858 (distribute_loop): Likewise.
8860 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8862 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
8863 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
8864 (p8_vmrgow_<mode>_direct): New define_insn.
8865 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
8866 handle endianness for vmrgew and vmrgow permute patterns.
8868 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
8870 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
8871 * config/rs6000/rs6000-cpus.def: Remove comment.
8872 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
8873 (POWERPC_MASKS): Likewise.
8874 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
8875 use of TARGET_VSX_TIMODE.
8876 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
8877 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
8878 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
8879 (rs6000_option_override_internal): Remove dead code.
8880 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
8881 (rs6000_legitimize_reload_address): Likewise.
8882 (rs6000_legitimate_address_p): Likewise.
8883 (rs6000_opt_masks): Delete "vsx-timode".
8884 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
8885 from function comment.
8886 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
8887 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
8888 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
8890 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
8891 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
8892 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
8895 (define_peephole2): Likewise.
8897 2017-08-17 Martin Sebor <msebor@redhat.com>
8900 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
8901 past the end of an array.
8902 (test_pp_format): Add test cases.
8904 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
8906 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
8907 missing ECF_NOTHROW flags.
8909 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
8912 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
8913 operands residing in integer registers.
8914 (*vsx_le_perm_load_<mode>): Likewise.
8915 (*vsx_le_perm_store_<mode>): Likewise.
8916 (define_peephole2): Add peepholes to optimize the above.
8918 2017-08-17 Marek Polacek <polacek@redhat.com>
8921 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
8922 to mimic what shorten_compare did. Change the return type to bool.
8923 (fold_cond_expr_with_comparison): Update call to
8924 operand_equal_for_comparison_p.
8925 (fold_ternary_loc): Likewise.
8927 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
8929 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
8931 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
8933 2017-08-17 Richard Biener <rguenther@suse.de>
8935 * tree-ssa-structalias.c (solve_graph): When propagating
8936 to successors update the graphs succ edges and avoid duplicate work.
8938 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
8941 * config/i386/i386.c (ix86_option_override_internal): Save target
8942 specific options after ix86_stack_protector_guard_reg was changed.
8944 2017-08-17 Richard Biener <rguenther@suse.de>
8946 PR tree-optimization/81827
8947 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
8948 (new_var_info): Initialize it conservatively.
8949 (get_call_vi): Mark register vars.
8950 (new_scalar_tmp_constraint_exp): Likewise.
8951 (handle_rhs_call): Likewise.
8952 (handle_const_call): Likewise.
8953 (create_function_info_for): Likewise.
8954 (solve_constraints): Sort varinfos to separate register from
8955 non-register vars to pack points-to solution bitmaps during
8958 2017-08-17 Marek Polacek <polacek@redhat.com>
8960 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
8962 2017-08-17 Richard Biener <rguenther@suse.de>
8964 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
8965 to true when overflow is undefined and we saturated the result.
8967 2017-08-17 Alan Modra <amodra@gmail.com>
8970 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
8971 Don't use store multiple if only one reg needs saving.
8972 (interesting_frame_related_regno): New function.
8973 (rs6000_frame_related): Don't emit frame info for regs that
8975 (rs6000_emit_epilogue): Likewise.
8977 2017-08-16 Nathan Sidwell <nathan@acm.org>
8979 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
8980 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
8981 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
8982 * tree.c (free_lang_data_in_type): Use else-if chain. Always
8983 clear TYPE_LANG_1. Remove obsolete member-function stripping.
8984 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
8985 (verify_type): Adjust for TYPE_BINFO move.
8986 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
8987 process TYPE_BINFO directly.
8988 (hash_tree): Likewise.
8989 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
8991 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
8994 2017-08-16 David Malcolm <dmalcolm@redhat.com>
8996 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
8998 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
9001 * config/i386/i386.md (*anddi_1_btr): Change predicates of
9002 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
9003 Add ix86_binary_operator_ok to insn constraint.
9004 (*iordi_1_bts): Ditto.
9005 (*xordi_1_btc): Ditto.
9006 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
9007 Update corresponding peephole2 pattern.
9011 2017-08-16 Bin Cheng <bin.cheng@arm.com>
9013 PR tree-optimization/81832
9014 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
9015 copy loop header which has IFN_LOOP_DIST_ALIAS call.
9017 2017-08-16 Marek Polacek <polacek@redhat.com>
9020 * fold-const.c (fold_indirect_ref_1): Restore original behavior
9021 regarding size_zero_node.
9023 2017-08-16 Martin Liska <mliska@suse.cz>
9026 * config.gcc: Respect previously set extra_objs in case
9029 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
9031 PR tree-optimization/81835
9032 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
9033 the phi SSA_NAME. Check that the condition in a COND_EXPR does
9034 not depend on the phi.
9036 2017-08-16 Alan Modra <amodra@gmail.com>
9038 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
9041 2017-08-16 Alan Modra <amodra@gmail.com>
9043 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
9044 (save_reg_p): ..into this. Update all callers.
9045 (first_reg_to_save): Simplify.
9047 2017-08-16 Alan Modra <amodra@gmail.com>
9049 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
9052 2017-08-15 Joseph Myers <joseph@codesourcery.com>
9056 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
9057 constant count if that count is less than 32.
9059 2017-08-15 Nathan Sidwell <nathan@acm.org>
9061 * gcc.c (execute): Emit friendlier message if inferior is killed
9062 by an external cause.
9064 2017-08-15 Richard Biener <rguenther@suse.de>
9066 PR tree-optimization/81790
9067 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
9068 CONSTRUCTORs from simplifying and VN.
9070 2017-08-14 Martin Sebor <msebor@redhat.com>
9072 * builtin-attrs.def: Add comments.
9074 2017-08-14 Martin Sebor <msebor@redhat.com>
9077 * doc/extend.texi (attribute nonstring): Document new attribute.
9079 2017-08-14 Martin Sebor <msebor@redhat.com>
9082 * tree-diagnostic.c (default_tree_printer): Handle %G.
9083 * gimple-pretty-print.h (percent_G_format): Declare new function.
9084 * gimple-pretty-print.c (percent_G_format): Define.
9085 * tree-pretty-print.c (percent_K_format): Add argument.
9087 2017-08-14 Martin Sebor <msebor@redhat.com>
9089 PR translation/79998
9090 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
9091 Remove a stray space.
9093 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
9096 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
9097 (*iordi_1_bts): Ditto.
9098 (*xordi_1_btc): Ditto.
9100 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9103 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
9104 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9106 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
9107 quoted strings, and make more translator-friendly.
9108 (darwin_rs6000_override_options): Likewise.
9109 (rs6000_option_override_internal): Likewise.
9110 (rs6000_return_in_memory): Fix overlong line.
9111 (init_cmulative_args): Use quoted strings, and make more
9112 translator-friendly.
9113 (rs6000_pass_by_reference): Fix overlong line.
9114 (def_builtin): Use quoted strings.
9115 (altivec_expand_predicate_builtin): Use quoted strings, and make
9116 more translator-friendly.
9117 (htm_expand_builtin): Use quoted strings.
9118 (cpu_expand_builtin): Use quoted strings, and make more
9119 translator-friendly.
9120 (altivec_expand_builtin): Likewise.
9121 (paired_expand_predicate_builtin): Likewise.
9122 (rs6000_invalid_builtin): Likewise.
9123 (builtin_function_type): Use quoted strings.
9124 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
9125 more translator-friendly.
9126 (rs6000_trampoline_init): Likewise.
9127 (rs6000_handle_altivec_attribute): Likewise.
9128 (rs6000_inner_target_options): Use quoted strings.
9129 (rs6000_disable_incompatible_switches): Likewise.
9130 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
9131 strings, and make more translator-friendly.
9132 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
9134 2017-08-14 Bin Cheng <bin.cheng@arm.com>
9136 PR tree-optimization/81799
9137 * tree-loop-distribution.c (version_loop_by_alias_check): Force
9138 cond_expr to simple gimple operand.
9140 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
9143 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
9145 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
9149 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
9150 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
9151 TARGET_GASISR_PROLOGUES.
9152 * config/avr/avr.c (avr_option_override): Same.
9153 (avr_pass_pre_proep::execute): Same.
9155 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
9158 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
9159 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
9161 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
9163 * config/i386/i386.md (*load_tp_<mode>): Redefine as
9164 define_insn_and_split. Split to a memory load from 0 in
9165 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
9166 using PTR mode iterator.
9167 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
9168 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
9169 (*add_tp_<mode>): Redefine as define_insn_and_split.
9170 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
9171 address space. Merge with *add_tp_x32 using PTR mode iterator.
9172 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
9173 Split to an add with a memory load from 0 in
9174 DEFAULT_TLS_SEG_REG address space.
9176 2017-08-12 Andrew Pinski <apinski@cavium.com>
9178 * config/aarch64/aarch64-option-extensions.def (rdma):
9179 Fix feature string to what Linux prints out in /proc/cpuinfo.
9181 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
9184 * dwarf2out.c (modified_type_die): For C typedef types that have
9185 an ultimate origin, process the ultimate origin instead of the
9187 (gen_typedef_die): Assert that input DECLs have no ultimate
9189 (gen_type_die_with_usage): For typedef variants that have an
9190 ultimate origin, just call gen_decl_die on the original DECL.
9191 (process_scope_var): Avoid creating DIEs for local typedefs and
9192 concrete static variables.
9194 2017-08-12 Alan Modra <amodra@gmail.com>
9198 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
9199 match gnu-user.h startfile.
9200 (ENDFILE_LINUX_SPEC): Similarly.
9202 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
9205 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
9207 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
9209 2017-08-11 Tamar Christina <tamar.christina@arm.com>
9210 * config/aarch64/aarch64.md (mov<mode>): Change.
9211 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
9212 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
9213 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
9215 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
9217 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
9218 for storage order barriers.
9220 2017-08-11 Martin Liska <mliska@suse.cz>
9223 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
9224 variables of void type.
9226 2017-08-11 Martin Liska <mliska@suse.cz>
9228 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
9229 TARGET_SUPPORTS_ALIASES.
9230 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
9231 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
9232 (optimize_weakref): Likewise.
9233 * symtab.c (symtab_node::noninterposable_alias): Likewise.
9234 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
9235 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
9237 2017-08-11 Martin Liska <mliska@suse.cz>
9240 * config/i386/i386.c (make_resolver_func): Do complete
9241 refactoring of the function.
9243 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
9246 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
9247 * config/i386/i386.c (ix86_stack_protect_guard): Use
9248 ix86_stack_protect_guard_symbol_str to generate varible declaration.
9249 * doc/invoke.texi (x86 Options): Document
9250 -mstack-protector-guard-symbol= option.
9252 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
9254 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
9255 * config/i386/i386.c (ix86_split_stack_guard): New function.
9256 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
9257 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
9258 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
9259 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
9260 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
9261 (split_stack_space_check): Call ix86_split_stack_guard.
9263 2017-08-10 Martin Sebor <msebor@redhat.com>
9265 * print-tree.c (print_node): Print location using the established
9267 Replace spaces with colons.
9268 (debug_raw, debug): Ditto.
9270 2017-08-10 Martin Sebor <msebor@redhat.com>
9273 * pretty-print.c (pp_format): Correct the handling of %s precision.
9275 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
9278 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
9280 (ix86_finalize_stack_frame_flags): This. Also clear
9281 frame_pointer_needed if -fno-omit-frame-pointer is used without
9283 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
9284 with ix86_finalize_stack_frame_flags.
9285 (ix86_expand_epilogue): Likewise.
9286 (ix86_expand_split_stack_prologue): Likewise.
9287 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
9289 2017-08-10 Martin Liska <mliska@suse.cz>
9292 * c-attribs.c (handle_target_attribute):
9293 Report warning for an empty string argument of target attribute.
9295 2017-08-09 Jakub Jelinek <jakub@redhat.com>
9298 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
9300 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
9301 or DECL_NONLOCAL labels.
9302 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
9303 or DECL_NONLOCAL labels here.
9305 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
9307 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
9308 to indicate when early gimple folding has been disabled.
9309 (rs6000_gimple_fold_builtin): Add debug content.
9310 (rs6000_invalid_builtin): Fix whitespace.
9311 (rs6000_expand_builtin): Fix whitespace.
9312 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
9314 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
9317 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
9318 SAVE_MULTIPLE if not all the registers that saves, should be saved.
9320 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
9322 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
9323 (qdf24xx): Likewise.
9324 * config/aarch64/aarch64.md: Include falkor.md.
9325 * config/aarch64/falkor.md: New.
9327 2017-08-09 Marek Polacek <polacek@redhat.com>
9330 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
9331 * diagnostic.c (emit_diagnostic): Add a comment.
9332 (emit_diagnostic_valist): New function.
9334 2017-08-09 Marek Polacek <polacek@redhat.com>
9337 * input.c (make_location): New overload.
9338 * input.h (make_location): Declare.
9340 2017-08-08 Alan Modra <amodra@gmail.com>
9341 H.J. Lu <hongjiu.lu@intel.com>
9344 * gcc.c (NO_PIE_SPEC): Delete.
9345 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
9347 (LINK_PIE_SPEC): ..to here.
9348 (LINK_COMMAND_SPEC): Support -no-pie.
9349 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
9350 chain of crtbegin*.o selection, update for PIE_SPEC changes and
9352 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
9353 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
9354 (ENDFILE_CRTEND_SPEC): Similarly.
9356 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
9359 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
9360 (mstack-protector-guard-offset=): Ditto.
9361 * config/i386/i386.c (ix86_option_override): Handle
9362 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
9364 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
9365 ix86_stack_protect_guard_offset variables.
9366 (TARGET_STACK_PROTECT_GUARD): Always define.
9367 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
9368 and -mstack-protector-guard-offset= options.
9370 2017-08-08 Bin Cheng <bin.cheng@arm.com>
9372 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
9373 boundary case for the last candidate.
9375 2017-08-08 Bin Cheng <bin.cheng@arm.com>
9377 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
9378 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
9380 2017-08-08 Tamar Christina <tamar.christina@arm.com>
9383 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
9384 * config/aarch64/aarch64-builtins.c
9385 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
9386 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
9387 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
9389 2017-08-08 Tamar Christina <tamar.christina@arm.com>
9390 Andrew Pinski <pinskia@gmail.com>
9393 * internal-fn.def (XORSIGN): New.
9394 * optabs.def (xorsign_optab): New.
9395 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
9396 (convert_expand_mult_copysign): New.
9397 (pass_optimize_widening_mul::execute): Call
9398 convert_expand_mult_copysign.
9400 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9402 PR tree-optimization/81354
9403 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
9404 Insert on edges rather than explicitly creating landing pads.
9405 (analyze_candidates_and_replace): Commit edge inserts.
9407 2017-08-08 Richard Biener <rguenther@suse.de>
9410 * tree-ssa-loop-niter.c: Include tree-dfa.h.
9411 (expand_simple_operations): Also look through ADDR_EXPRs with
9412 MEM_REF bases treating them as POINTER_PLUS_EXPR.
9414 2017-08-08 Richard Biener <rguenther@suse.de>
9416 PR tree-optimization/81723
9417 * tree-vect-slp.c (struct bst_traits): New hash traits.
9418 (bst_fail): New global.
9419 (vect_build_slp_tree_2): New worker, split out from ...
9420 (vect_build_slp_tree): ... this now wrapping it with using
9421 bst_fail set to cache SLP tree build fails. Properly handle
9423 (vect_analyze_slp_instance): Allocate and free bst_fail.
9425 2017-08-08 Martin Liska <mliska@suse.cz>
9428 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
9429 LABEL_DECLs that can be from a different function.
9431 2017-08-08 Bin Cheng <bin.cheng@arm.com>
9433 PR tree-optimization/81744
9434 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
9435 loop's number of iterations.
9437 2017-08-08 Martin Liska <mliska@suse.cz>
9439 * asan.c: Include header files.
9440 * attribs.c (build_decl_attribute_variant): New function moved
9442 (build_type_attribute_qual_variant): Likewise.
9443 (cmp_attrib_identifiers): Likewise.
9444 (simple_cst_list_equal): Likewise.
9445 (omp_declare_simd_clauses_equal): Likewise.
9446 (attribute_value_equal): Likewise.
9447 (comp_type_attributes): Likewise.
9448 (build_type_attribute_variant): Likewise.
9449 (lookup_ident_attribute): Likewise.
9450 (remove_attribute): Likewise.
9451 (merge_attributes): Likewise.
9452 (merge_type_attributes): Likewise.
9453 (merge_decl_attributes): Likewise.
9454 (merge_dllimport_decl_attributes): Likewise.
9455 (handle_dll_attribute): Likewise.
9456 (attribute_list_equal): Likewise.
9457 (attribute_list_contained): Likewise.
9458 * attribs.h (lookup_attribute): New function moved from tree.[ch].
9459 (lookup_attribute_by_prefix): Likewise.
9460 * bb-reorder.c: Include header files.
9461 * builtins.c: Likewise.
9462 * calls.c: Likewise.
9463 * cfgexpand.c: Likewise.
9464 * cgraph.c: Likewise.
9465 * cgraphunit.c: Likewise.
9466 * convert.c: Likewise.
9467 * dwarf2out.c: Likewise.
9468 * final.c: Likewise.
9469 * fold-const.c: Likewise.
9470 * function.c: Likewise.
9471 * gimple-expr.c: Likewise.
9472 * gimple-fold.c: Likewise.
9473 * gimple-pretty-print.c: Likewise.
9474 * gimple.c: Likewise.
9475 * gimplify.c: Likewise.
9476 * hsa-common.c: Likewise.
9477 * hsa-gen.c: Likewise.
9478 * internal-fn.c: Likewise.
9479 * ipa-chkp.c: Likewise.
9480 * ipa-cp.c: Likewise.
9481 * ipa-devirt.c: Likewise.
9482 * ipa-fnsummary.c: Likewise.
9483 * ipa-inline.c: Likewise.
9484 * ipa-visibility.c: Likewise.
9486 * lto-cgraph.c: Likewise.
9487 * omp-expand.c: Likewise.
9488 * omp-general.c: Likewise.
9489 * omp-low.c: Likewise.
9490 * omp-offload.c: Likewise.
9491 * omp-simd-clone.c: Likewise.
9492 * opts-global.c: Likewise.
9493 * passes.c: Likewise.
9494 * predict.c: Likewise.
9495 * sancov.c: Likewise.
9496 * sanopt.c: Likewise.
9497 * symtab.c: Likewise.
9498 * toplev.c: Likewise.
9499 * trans-mem.c: Likewise.
9500 * tree-chkp.c: Likewise.
9501 * tree-eh.c: Likewise.
9502 * tree-into-ssa.c: Likewise.
9503 * tree-object-size.c: Likewise.
9504 * tree-parloops.c: Likewise.
9505 * tree-profile.c: Likewise.
9506 * tree-ssa-ccp.c: Likewise.
9507 * tree-ssa-live.c: Likewise.
9508 * tree-ssa-loop.c: Likewise.
9509 * tree-ssa-sccvn.c: Likewise.
9510 * tree-ssa-structalias.c: Likewise.
9511 * tree-ssa.c: Likewise.
9512 * tree-streamer-in.c: Likewise.
9513 * tree-vectorizer.c: Likewise.
9514 * tree-vrp.c: Likewise.
9516 * ubsan.c: Likewise.
9517 * varasm.c: Likewise.
9518 * varpool.c: Likewise.
9519 * tree.c: Remove functions moved to attribs.[ch].
9521 * config/aarch64/aarch64.c: Add attrs.h header file.
9522 * config/alpha/alpha.c: Likewise.
9523 * config/arc/arc.c: Likewise.
9524 * config/arm/arm.c: Likewise.
9525 * config/avr/avr.c: Likewise.
9526 * config/bfin/bfin.c: Likewise.
9527 * config/c6x/c6x.c: Likewise.
9528 * config/cr16/cr16.c: Likewise.
9529 * config/cris/cris.c: Likewise.
9530 * config/darwin.c: Likewise.
9531 * config/epiphany/epiphany.c: Likewise.
9532 * config/fr30/fr30.c: Likewise.
9533 * config/frv/frv.c: Likewise.
9534 * config/ft32/ft32.c: Likewise.
9535 * config/h8300/h8300.c: Likewise.
9536 * config/i386/winnt.c: Likewise.
9537 * config/ia64/ia64.c: Likewise.
9538 * config/iq2000/iq2000.c: Likewise.
9539 * config/lm32/lm32.c: Likewise.
9540 * config/m32c/m32c.c: Likewise.
9541 * config/m32r/m32r.c: Likewise.
9542 * config/m68k/m68k.c: Likewise.
9543 * config/mcore/mcore.c: Likewise.
9544 * config/microblaze/microblaze.c: Likewise.
9545 * config/mips/mips.c: Likewise.
9546 * config/mmix/mmix.c: Likewise.
9547 * config/mn10300/mn10300.c: Likewise.
9548 * config/moxie/moxie.c: Likewise.
9549 * config/msp430/msp430.c: Likewise.
9550 * config/nds32/nds32-isr.c: Likewise.
9551 * config/nds32/nds32.c: Likewise.
9552 * config/nios2/nios2.c: Likewise.
9553 * config/nvptx/nvptx.c: Likewise.
9554 * config/pa/pa.c: Likewise.
9555 * config/pdp11/pdp11.c: Likewise.
9556 * config/powerpcspe/powerpcspe.c: Likewise.
9557 * config/riscv/riscv.c: Likewise.
9558 * config/rl78/rl78.c: Likewise.
9559 * config/rx/rx.c: Likewise.
9560 * config/s390/s390.c: Likewise.
9561 * config/sh/sh.c: Likewise.
9562 * config/sol2.c: Likewise.
9563 * config/sparc/sparc.c: Likewise.
9564 * config/spu/spu.c: Likewise.
9565 * config/stormy16/stormy16.c: Likewise.
9566 * config/tilegx/tilegx.c: Likewise.
9567 * config/tilepro/tilepro.c: Likewise.
9568 * config/v850/v850.c: Likewise.
9569 * config/vax/vax.c: Likewise.
9570 * config/visium/visium.c: Likewise.
9571 * config/xtensa/xtensa.c: Likewise.
9573 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9576 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
9577 constraints since the -mupper-regs-* switches have been
9579 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
9580 into a vector from a double word element that was extracted from
9581 another vector, and eliminate extra XXPERMDI instructions.
9582 (vsx_concat_<mode>_2): Likewise.
9583 (vsx_concat_<mode>_3): Likewise.
9584 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
9585 concat to allow optimizing inserts from previous extracts.
9587 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
9589 * config/i386/i386.c (ix86_stack_protect_guard): Generate
9590 memory reference to a SSP offset in TLS address space.
9591 (ix86_print_operand) <case '@'>: Remove.
9592 (ix86_print_operand_punct_valid_p): Remove '@' code.
9593 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
9595 (stack_tls_protect_set_<mode>): Remove.
9596 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
9597 (stack_tls_protect_test_<mode>): Remove.
9598 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
9600 2017-08-07 Olivier Hainque <hainque@adacore.com>
9603 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
9605 2017-08-07 Douglas Rupp <rupp@adacore.com>
9607 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
9608 variable was referenced as multidir in command.
9610 2017-08-07 Jakub Jelinek <jakub@redhat.com>
9613 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
9616 2017-08-07 Martin Liska <mliska@suse.cz>
9618 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
9619 * config/rl78/rl78.c: Add include of attribs.h.
9620 * config/sh/sh.c: Likewise.
9621 * config/v850/v850.c: Likewise.
9623 2017-08-07 Tom de Vries <tom@codesourcery.com>
9626 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
9628 2017-08-07 Martin Liska <mliska@suse.cz>
9630 * config/mips/mips.c: Include attribs.h.
9632 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
9635 * doc/invoke.texi: Document change in behvaior for -Ofast for
9638 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
9640 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
9642 (aarch64_pop_regs): Likewise.
9643 (aarch64_gen_load_pair): Likewise.
9644 (aarch64_save_callee_saves): Likewise.
9645 (aarch64_restore_callee_saves): Likewise.
9647 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
9649 * config/i386/i386.c: Revert the last change.
9651 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
9654 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
9656 (ix86_finalize_stack_frame_flags): This. Also clear
9657 frame_pointer_needed if -fno-omit-frame-pointer is used without
9659 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
9660 with ix86_finalize_stack_frame_flags.
9661 (ix86_expand_epilogue): Likewise.
9662 (ix86_expand_split_stack_prologue): Likewise.
9664 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
9667 * config/i386/i386.c (get_builtin_code_for_version): Set priority
9668 to P_AES for Westmere.
9670 2017-08-07 Jonathan Yong <10walls@gmail.com>
9672 * config/i386/mingw.opt (fset-stack-executable): Removed.
9673 * config/i386/cygming.opt (fset-stack-executable): Moved
9675 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
9677 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
9679 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
9681 2017-08-07 Marek Polacek <polacek@redhat.com>
9684 * fold-const.c (fold_indirect_ref_1): Check type_domain.
9686 2017-08-07 Martin Liska <mliska@suse.cz>
9688 * attribs.h (canonicalize_attr_name): New function.
9689 (cmp_attribs): Move from c-format.c and adjusted.
9690 (is_attribute_p): Moved from tree.h.
9691 * tree-inline.c: Add new includes.
9692 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
9693 (private_is_attribute_p): Remove.
9694 (private_lookup_attribute): Likewise.
9695 (private_lookup_attribute_by_prefix): Simplify.
9696 (remove_attribute): Use is_attribute_p.
9697 * tree.h: Remove removed declarations.
9699 2017-08-07 Jakub Jelinek <jakub@redhat.com>
9702 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
9703 instead of computing it in the function. Formatting fix.
9704 (expand_case): Don't rely on default_edge being the first edge,
9705 clear it if removing it, pass default_edge to
9706 emit_case_dispatch_table.
9707 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
9710 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
9712 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
9713 insn in the function, emit NOP after the insn.
9715 2017-08-06 Tom de Vries <tom@codesourcery.com>
9717 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
9720 2017-08-06 Tom de Vries <tom@codesourcery.com>
9722 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
9725 2017-08-04 Yury Gribov <tetra2005@gmail.com>
9727 PR tree-optimization/57371
9728 * match.pd: New pattern.
9730 2017-08-04 Marek Polacek <polacek@redhat.com>
9733 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
9734 perform the computation in offset_int.
9736 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
9738 PR tree-optimization/81136
9739 * tree-vectorizer.h: Include tree-hash-traits.h.
9740 (vec_base_alignments): New typedef.
9741 (vec_info): Add a base_alignments field.
9742 (vect_record_base_alignments): Declare.
9743 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
9745 (DR_IS_CONDITIONAL_IN_STMT): New macro.
9746 (create_data_ref): Add an is_conditional_in_stmt argument.
9747 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
9748 the is_conditional_in_stmt field.
9749 (data_ref_loc): Add an is_conditional_in_stmt field.
9750 (get_references_in_stmt): Set the is_conditional_in_stmt field.
9751 (find_data_references_in_stmt): Update call to create_data_ref.
9752 (graphite_find_data_references_in_stmt): Likewise.
9753 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
9754 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
9755 (vect_record_base_alignment): New function.
9756 (vect_record_base_alignments): Likewise.
9757 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
9758 for nested statements even if we fail to compute a misalignment.
9759 Use pooled base alignments for unconditional references.
9760 (vect_find_same_alignment_drs): Compare base addresses instead
9762 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
9763 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
9765 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
9767 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
9768 Add an explicit name for the enum. Use auto_vec for slp_instances
9770 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
9772 (_bb_vec_info): Add a constructor and destructor.
9773 (vinfo_for_stmt): Return NULL for uids of -1 as well.
9774 (destroy_loop_vec_info): Delete.
9775 (vect_destroy_datarefs): Likewise.
9776 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
9777 (vec_info::vec_info): New function.
9778 (vec_info::~vec_info): Likewise.
9779 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
9780 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
9781 destroy_loop_vec_info.
9782 * tree-vect-loop.c (new_loop_vec_info): Replace with...
9783 (_loop_vec_info::_loop_vec_info): ...this.
9784 (destroy_loop_vec_info): Replace with...
9785 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
9786 the stmt_vec_infos. Leave handling of vec_info information to its
9787 destructor. Remove explicit vector releases.
9788 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
9789 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
9790 * tree-vect-slp.c (new_bb_vec_info): Replace with...
9791 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
9792 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
9793 (destroy_bb_vec_info): Replace with...
9794 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
9795 information to its destructor.
9796 (vect_slp_analyze_bb_1): Use new and delete instead of
9797 new_bb_vec_info and destroy_bb_vec_info.
9798 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
9801 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
9803 * tree-data-ref.h (subscript): Add access_fn field.
9804 (data_dependence_relation): Add could_be_independent_p.
9805 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
9806 (same_access_functions): Move to tree-data-ref.c.
9807 * tree-data-ref.c (ref_contains_union_access_p): New function.
9808 (access_fn_component_p): Likewise.
9809 (access_fn_components_comparable_p): Likewise.
9810 (dr_analyze_indices): Add a reference to access_fn_component_p.
9811 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
9813 (constant_access_functions): Likewise.
9814 (add_other_self_distances): Likewise.
9815 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
9816 (initialize_data_dependence_relation): Use XCNEW and remove
9817 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
9818 of access functions that have the same type. Allow the
9819 subsequence to end with different bases in some circumstances.
9820 Record the chosen access functions in SUB_ACCESS_FN.
9821 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
9822 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
9823 (subscript_dependence_tester_1): Likewise dra and drb.
9824 (build_classic_dist_vector): Update calls accordingly.
9825 (subscript_dependence_tester): Likewise.
9826 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
9827 DDR_COULD_BE_INDEPENDENT_P.
9828 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
9829 comp_alias_ddrs instead of may_alias_ddrs.
9830 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
9832 (vect_analyze_data_ref_dependence): Use it if
9833 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
9834 distance vectors if that fails.
9835 (dependence_distance_ge_vf): New function.
9836 (vect_prune_runtime_alias_test_list): Use it. Don't clear
9837 LOOP_VINFO_MAY_ALIAS_DDRS.
9839 2017-08-04 Richard Biener <rguenther@suse.de>
9842 * fold-const.c (fold_binary_loc): Properly restrict
9843 minus_var0 && minus_var1 case when associating undefined overflow
9846 2017-08-04 Tom de Vries <tom@codesourcery.com>
9848 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
9850 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9852 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9853 Don't start diagnostic messages with a capital letter.
9854 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9856 (rs6000_invalid_builtin): Likewise.
9857 (rs6000_trampoline_init): Likewise.
9859 2017-08-03 Jakub Jelinek <jakub@redhat.com>
9862 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
9863 after setting changeable df flags.
9865 2017-08-03 Richard Biener <rguenther@suse.de>
9867 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
9868 up if the use is in USE - X.
9870 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
9872 * toplev.c (dumpfile.h): New include.
9873 (internal_error_reentered): New static function. Use it...
9874 (internal_error_function): ...here to handle reentered internal_error.
9876 2017-08-03 Richard Biener <rguenther@suse.de>
9879 * fold-const.c (split_tree): Add minus_var and minus_con
9880 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
9881 here but always use minus_*.
9882 (associate_trees): Assert we never associate with MINUS_EXPR
9883 and NULL first operand. Do not recurse for PLUS_EXPR operands
9884 when associating as MINUS_EXPR either.
9885 (fold_binary_loc): Track minus_var and minus_con.
9887 2017-08-03 Tom de Vries <tom@codesourcery.com>
9890 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
9891 ACCEL_COMPILER, apply finish_options on
9892 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
9894 2017-08-03 Tom de Vries <tom@codesourcery.com>
9897 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
9898 function_entry_patch_area_size > 0.
9900 2017-08-03 Jakub Jelinek <jakub@redhat.com>
9903 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
9904 instead of 10??LU, perform unit multiplication in wide_int,
9905 don't change alloc_object_size_limit if the limit is larger
9908 PR tree-optimization/81655
9909 PR tree-optimization/81588
9910 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
9911 the case when ranges[i].low and high are 1 for unsigned type with
9915 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
9916 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
9918 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9920 * tree-vrp.h: Add include guard.
9922 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
9925 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
9926 (ud2): New insn pattern.
9927 * config/i386/i386.c (ix86_expand_epilogue):
9928 For naked functions, generate ud2 instead of trap insn.
9930 2017-08-02 Marek Polacek <polacek@redhat.com>
9933 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
9935 2017-08-02 Tom de Vries <tom@codesourcery.com>
9936 Cesar Philippidis <cesar@codesourcery.com>
9938 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
9939 Add missing edge probabilities.
9941 2017-08-02 Tamar Christina <tamar.christina@arm.com>
9943 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
9946 2017-08-02 Jakub Jelinek <jakub@redhat.com>
9949 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
9950 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
9951 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
9952 prologue_seq sequences - if any.
9954 2017-08-02 Richard Biener <rguenther@suse.de>
9956 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
9957 via vectors if supported, integer extracts via punning if supported
9958 or otherwise vector extracts.
9960 2017-08-02 Richard Biener <rguenther@suse.de>
9962 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
9964 (bitmap_insert_into_set): ... this.
9966 2017-08-02 Richard Biener <rguenther@suse.de>
9968 PR tree-optimization/81633
9970 2015-08-17 Alan Hayward <alan.hayward@arm.com>
9972 PR tree-optimization/71752
9973 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
9975 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
9977 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
9978 (machine_function::call_ms2sysv_pad_out): Remove field.
9979 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
9980 (ix86_compute_frame_layout): Likewise.
9982 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
9985 * config/i386/i386.c (ix86_set_func_type): Disallow naked
9986 attribute with interrupt attribute.
9988 2017-08-01 Andrew Pinski <apinski@cavium.com>
9990 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
9991 BIT_INSERT_EXPR's operand 1
9992 to see if the types precision matches.
9994 2017-08-01 Martin Liska <mliska@suse.cz>
9997 * builtins.c (expand_builtin_memcpy_args): Remove.
9998 (expand_builtin_memcpy): Call newly added function
9999 expand_builtin_memory_copy_args.
10000 (expand_builtin_memcpy_with_bounds): Likewise.
10001 (expand_builtin_mempcpy): Remove last argument.
10002 (expand_builtin_mempcpy_with_bounds): Likewise.
10003 (expand_builtin_memory_copy_args): New function created from
10004 expand_builtin_mempcpy_args with small modifications.
10005 (expand_builtin_mempcpy_args): Remove.
10006 (expand_builtin_stpcpy): Remove unused argument.
10007 (expand_builtin): Likewise.
10008 (expand_builtin_with_bounds): Likewise.
10010 2017-08-01 Martin Liska <mliska@suse.cz>
10013 Make mempcpy more optimal (PR middle-end/70140).
10015 2017-08-01 Jakub Jelinek <jakub@redhat.com>
10018 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
10019 __builtin_vec_cmpne verify both arguments are compatible vectors
10020 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
10021 verify arg1_type is a pointer or array type. For __builtin_vec_st,
10022 move computation of aligned to after checking the argument types.
10026 * config/rs6000/vsx.md (vextract_fp_from_shorth,
10027 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
10030 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
10031 Doug Rupp <rupp@adacore.com>
10032 Olivier Hainque <hainque@adacore.com>
10034 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
10035 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
10037 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
10038 for TARGET_OS_CPP_BUILTIN.
10039 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
10040 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
10042 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
10043 passing required abi options to the assembler for EABI configurations.
10044 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
10045 of .text.hot and .text.unlikely sections for kernel modules when
10046 using ARM style exceptions.
10047 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
10048 options. Add EXTRA_CC1_SPEC.
10049 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
10051 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
10053 (ARM_TARGET2_DWARF_FORMAT): Define.
10054 * config/arm/t-vxworks: Adjust multilib control to removal of the
10055 Diab command line options.
10057 2017-08-01 Martin Liska <mliska@suse.cz>
10059 PR gcov-profile/81561
10060 * gcov.c (unblock): Make unblocking safe as we need to preserve
10061 index correspondence of blocks and block_lists.
10063 2017-08-01 Richard Biener <rguenther@suse.de>
10065 PR tree-optimization/81181
10066 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10067 (compute_antic): ... end of iteration here.
10069 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
10071 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
10072 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
10073 (ftree-slp-vectorize): Likewise.
10074 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
10075 can no longer be set independent of flag_tree_loop_vectorize.
10076 * omp-general.c (emp_max_vf): Likewise.
10077 * opts.c (enable_fdo_optimizations): Remove references to
10078 flag_tree_vectorize, these are now implicit.
10079 (common_handle_option): Remove handling for OPT_ftree_vectorize,
10080 and leave it for the options machinery.
10082 2017-08-01 Martin Liska <mliska@suse.cz>
10084 PR middle-end/70140
10085 * builtins.c (expand_builtin_memcpy_args): Remove.
10086 (expand_builtin_memcpy): Call newly added function
10087 expand_builtin_memory_copy_args.
10088 (expand_builtin_memcpy_with_bounds): Likewise.
10089 (expand_builtin_mempcpy): Remove last argument.
10090 (expand_builtin_mempcpy_with_bounds): Likewise.
10091 (expand_builtin_memory_copy_args): New function created from
10092 expand_builtin_mempcpy_args with small modifications.
10093 (expand_builtin_mempcpy_args): Remove.
10094 (expand_builtin_stpcpy): Remove unused argument.
10095 (expand_builtin): Likewise.
10096 (expand_builtin_with_bounds): Likewise.
10098 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
10101 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
10102 print "ds:" only for immediates in generic address space.
10104 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
10107 * config/i386/i386.c (ix86_funciton_naked): New prototype.
10108 (ix86_function_ok_for_sibcall): Return false for naked functions.
10110 2017-08-01 Richard Biener <rguenther@suse.de>
10112 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
10113 (compute_antic): Seed worklist with exit block predecessors.
10114 * cfganal.c (dfs_find_deadend): For a cycle return the source
10115 of the edge closing it.
10117 2017-08-01 Tamar Christina <tamar.christina@arm.com>
10119 * config/aarch64/aarch64.c
10120 (aarch64_can_const_movi_rtx_p): Move 0 check.
10122 2017-08-01 Bin Cheng <bin.cheng@arm.com>
10124 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
10125 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
10127 * match.pd: Ditto in address comparison pattern.
10129 2017-08-01 Bin Cheng <bin.cheng@arm.com>
10131 PR tree-optimization/81627
10132 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
10133 closed ssa form for store-store chain.
10135 2017-08-01 Bin Cheng <bin.cheng@arm.com>
10137 PR tree-optimization/81620
10138 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
10139 for store-store chain.
10141 2017-08-01 Jakub Jelinek <jakub@redhat.com>
10143 PR tree-optimization/81588
10144 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
10145 ranges[i].in_p, invert comparison code ccode. For >/>=,
10146 swap rhs1 and rhs2 and comparison code unconditionally,
10147 for </<= don't do that. Don't swap rhs1/rhs2 again if
10148 ranges[i].in_p, instead invert comparison code ccode if
10149 opcode or oe->rank is BIT_IOR_EXPR.
10152 * optabs.def (vec_extract_optab, vec_init_optab): Change from
10153 a direct optab to conversion optab.
10154 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
10155 with GET_MODE_INNER as last argument instead of optab_handler.
10156 * expmed.c (extract_bit_field_1): Likewise. Use vector from
10157 vector extraction if possible and optab is available.
10158 * expr.c (store_constructor): Use convert_optab_handler instead
10159 of optab_handler. Use vector initialization from smaller
10160 vectors if possible and optab is available.
10161 * tree-vect-stmts.c (vectorizable_load): Likewise.
10162 * doc/md.texi (vec_extract, vec_init): Document that the optabs
10163 now have two modes.
10164 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
10165 of vec_init from half-sized vectors with the same element mode.
10166 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
10167 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
10168 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
10169 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
10170 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
10171 after mode in gen_vec_extract* calls.
10172 (vec_extract<mode>): Renamed to ...
10173 (vec_extract<mode><ssescalarmodelower>): ... this.
10174 (vec_extract<mode><ssehalfvecmodelower>): New expander.
10175 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
10176 element mode after mode in gen_vec_init* calls.
10177 (VEC_INIT_HALF_MODE): New mode iterator.
10178 (vec_init<mode>): Renamed to ...
10179 (vec_init<mode><ssescalarmodelower>): ... this.
10180 (vec_init<mode><ssehalfvecmodelower>): New expander.
10181 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
10182 (vec_extractv2sfsf): ... this.
10183 (vec_initv2sf): Renamed to ...
10184 (vec_initv2sfsf): ... this.
10185 (vec_extractv2si): Renamed to ...
10186 (vec_extractv2sisi): ... this.
10187 (vec_initv2si): Renamed to ...
10188 (vec_initv2sisi): ... this.
10189 (vec_extractv4hi): Renamed to ...
10190 (vec_extractv4hihi): ... this.
10191 (vec_initv4hi): Renamed to ...
10192 (vec_initv4hihi): ... this.
10193 (vec_extractv8qi): Renamed to ...
10194 (vec_extractv8qiqi): ... this.
10195 (vec_initv8qi): Renamed to ...
10196 (vec_initv8qiqi): ... this.
10197 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
10198 (vec_init<mode>): Renamed to ...
10199 (vec_init<mode><VEC_base_l>): ... this.
10200 (vec_extract<mode>): Renamed to ...
10201 (vec_extract<mode><VEC_base_l>): ... this.
10202 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
10203 (vec_initv2sfsf): ... this.
10204 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
10205 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
10206 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
10207 element mode after mode in gen_vec_init* calls.
10208 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
10209 (vec_init<mode><Vel>): ... this.
10210 (vec_extract<mode>): Renamed to ...
10211 (vec_extract<mode><Vel>): ... this.
10212 * config/aarch64/iterators.md (Vel): New mode attribute.
10213 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
10214 Add element mode after mode in gen_vec_extract* calls.
10215 * config/s390/vector.md (non_vec_l): New mode attribute.
10216 (vec_extract<mode>): Renamed to ...
10217 (vec_extract<mode><non_vec_l>): ... this.
10218 (vec_init<mode>): Renamed to ...
10219 (vec_init<mode><non_vec_l>): ... this.
10220 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
10221 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
10223 * config/arm/iterators.md (V_elem_l): New mode attribute.
10224 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
10225 (vec_extract<mode><V_elem_l>): ... this.
10226 (vec_extractv2di): Renamed to ...
10227 (vec_extractv2didi): ... this.
10228 (vec_init<mode>): Renamed to ...
10229 (vec_init<mode><V_elem_l>): ... this.
10230 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
10231 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
10232 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
10233 Add element mode after gen_vec_extract* calls.
10234 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
10235 (vec_init<mode><unitmode>): ... this.
10236 (vec_extract<mode>): Renamed to ...
10237 (vec_extract<mode><unitmode>): ... this.
10238 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
10239 (vec_init<mode><unitmode>): ... this.
10240 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
10241 (vec_initv2sfsf): ... this.
10242 (vec_extractv2sf): Renamed to ...
10243 (vec_extractv2sfsf): ... this.
10244 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
10245 Add element mode after gen_vec_extract* calls.
10246 * config/mips/mips.md (unitmode): New mode iterator.
10247 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
10248 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
10249 * config/spu/spu.md (inner_l): New mode attribute.
10250 (vec_init<mode>): Renamed to ...
10251 (vec_init<mode><inner_l>): ... this.
10252 (vec_extract<mode>): Renamed to ...
10253 (vec_extract<mode><inner_l>): ... this.
10254 * config/sparc/sparc.md (veltmode): New mode iterator.
10255 (vec_init<VMALL:mode>): Renamed to ...
10256 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
10257 * config/ia64/vect.md (vec_initv2si): Renamed to ...
10258 (vec_initv2sisi): ... this.
10259 (vec_initv2sf): Renamed to ...
10260 (vec_initv2sfsf): ... this.
10261 (vec_extractv2sf): Renamed to ...
10262 (vec_extractv2sfsf): ... this.
10263 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
10264 (vec_init<mode>): Renamed to ...
10265 (vec_init<mode><VEC_base_l>): ... this.
10266 (vec_extract<mode>): Renamed to ...
10267 (vec_extract<mode><VEC_base_l>): ... this.
10268 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
10269 (vec_initv2sfsf): ... this.
10270 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
10271 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
10272 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
10273 gen_vec_init* calls.
10275 2017-08-01 Richard Biener <rguenther@suse.de>
10277 PR tree-optimization/81297
10278 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
10279 TREE_OVERFLOW from INTEGER_CSTs.
10281 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
10283 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
10285 2017-07-31 Carl Love <cel@us.ibm.com>
10287 * config/rs6000/rs6000-c: Add support for built-in functions
10288 vector signed char vec_xl_be (signed long long, signed char *);
10289 vector unsigned char vec_xl_be (signed long long, unsigned char *);
10290 vector signed int vec_xl_be (signed long long, signed int *);
10291 vector unsigned int vec_xl_be (signed long long, unsigned int *);
10292 vector signed long long vec_xl_be (signed long long, signed long long *);
10293 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
10294 vector signed short vec_xl_be (signed long long, signed short *);
10295 vector unsigned short vec_xl_be (signed long long, unsigned short *);
10296 vector double vec_xl_be (signed long long, double *);
10297 vector float vec_xl_be (signed long long, float *);
10298 * config/rs6000/altivec.h (vec_xl_be): Add #define.
10299 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
10300 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
10302 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
10303 (altivec_expand_builtin): Add switch statement to call
10304 altivec_expand_xl_be for each builtin.
10305 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
10306 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
10307 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
10308 __builtin_vsx_le_be_v16qi.
10309 * doc/extend.texi: Update the built-in documentation file for the
10310 new built-in functions.
10312 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
10315 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
10317 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
10319 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10321 * config.gcc: Add z14.
10322 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
10323 CPU model numbers for z13s and z14.
10324 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
10326 * config/s390/s390-opts.h (enum processor_type): Rename
10327 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
10328 * config/s390/s390.c (processor_table): Add field for CPU name to
10329 be passed to Binutils.
10330 (s390_asm_output_machine_for_arch): Use the new field in
10331 processor_table for Binutils.
10332 (s390_expand_builtin): Replace arch12 with z14.
10333 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
10334 (s390_get_sched_attrmask): Likewise.
10335 (s390_get_unit_mask): Likewise.
10336 * config/s390/s390.opt: Add z14 to processor_type enum.
10338 2017-07-31 Martin Jambor <mjambor@suse.cz>
10341 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
10342 regardless of optimization level.
10344 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
10345 Martin Liska <mliska@suse.cz>
10347 * predict.def: Remove old comment and adjust probability.
10348 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
10349 PREDICT statements.
10351 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
10354 * config/i386/i386.c (ix86_function_naked): New function.
10355 (ix86_can_use_return_insn_p): Return false for naked functions.
10356 (ix86_expand_prologue): Skip prologue for naked functions.
10357 (ix86_expand_epilogue): Skip epilogue for naked functions
10358 and emit trap instruction.
10359 (ix86_warn_func_return): New function.
10360 (ix86_attribute_table): Add "naked" attribute specification.
10361 (TARGET_WARN_FUNC_RETURN): Define.
10362 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
10364 2017-07-31 Martin Liska <mliska@suse.cz>
10366 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
10367 (dump_gimple_bb_header): Always dump BB info.
10368 (pp_cfg_jump): Do not append info about BB when dumping a jump.
10370 2017-07-31 Martin Liska <mliska@suse.cz>
10373 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
10374 also with current_function_decl non-null equality.
10376 2017-07-31 Jakub Jelinek <jakub@redhat.com>
10379 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
10380 change type to the element type, instead add eltype variable and
10381 use it where we are interested in the element type.
10383 PR tree-optimization/81603
10384 * ipa-polymorphic-call.c
10385 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
10386 offset arithmetic in offset_int, bail out if the resulting bit offset
10387 doesn't fit into shwi.
10389 2017-07-31 Martin Liska <mliska@suse.cz>
10391 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
10392 (gimplify_save_expr): Fix comment.
10394 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
10397 * config/i386/i386.c (ix86_function_arg): Update arguments for
10399 (ix86_compute_frame_layout): Set the initial stack offset to
10400 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
10401 INCOMING_FRAME_SP_OFFSET.
10402 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
10403 stack before exception handler returns.
10404 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
10405 the 'ERROR_CODE' for exception handler.
10407 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
10409 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
10410 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
10411 (ASM_OUTPUT_REG_POP): Ditto.
10412 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
10413 instead of asm_fprintf to output pure string.
10415 2017-07-29 Jakub Jelinek <jakub@redhat.com>
10417 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
10418 to imported_module_or_decl hook.
10419 (debug_nothing_tree_tree_tree_bool): Remove.
10420 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
10421 * debug.c (do_nothing_debug_hooks): Use
10422 debug_nothing_tree_tree_tree_bool_bool instead of
10423 debug_nothing_tree_tree_tree_bool.
10424 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
10425 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
10426 * sdbout.c (sdb_debug_hooks): Likewise.
10427 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
10428 (gen_namespace_die): Add DW_AT_export_symbols attribute if
10430 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
10431 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
10432 attribute, don't add anything.
10434 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10436 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
10437 (fold_build2_stat_loc): Likewise.
10438 (fold_build3_stat_loc): Likewise.
10439 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
10440 (fold_build1_loc): Remove macro.
10441 (fold_build2_loc): Likewise.
10442 (fold_build3_loc): Likewise.
10444 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10446 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
10447 (gimple_build_debug_bind_source_stat): Likewise.
10448 * gimple.h (gimple_build_debug_bind): Remove macro.
10449 (gimple_build_debug_bind_source): Likewise.
10451 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10453 * bitmap.c (bitmap_alloc): Adjust.
10454 (bitmap_gc_alloc): Likewise.
10455 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
10457 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10459 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
10460 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
10461 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
10462 (bitmap_gc_alloc_stat): Likewise.
10463 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
10465 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10467 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
10468 * rtl.h (shallow_copy_rtx): Remove macro.
10470 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10472 * emit-rtl.c (gen_raw_REG): Adjust.
10473 * gengenrtl.c (gendef): Likewise.
10474 * rtl.c (rtx_alloc_stat): Remove _stat from name.
10475 * rtl.h (rtx_alloc): Remove macro.
10477 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10479 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
10480 (build_tree_list_stat): Likewise.
10481 * tree.h (build_tree_list): Remove macro.
10482 (build_tree_list_vec): Likewise.
10484 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10486 * tree.c (make_vector_stat): Remove _stat from name.
10487 (build_vector_stat): Likewise.
10488 * tree.h (make_vector_stat): Remove macro.
10489 (build_vector_stat): Likewise.
10491 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10493 * tree.h (build_var_debug_value): Remove prototype.
10495 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10497 * tree.c (tree_cons_stat): Remove _stat from name.
10498 * tree.h (tree_cons): Remove macro.
10500 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10502 * tree.c (build_vl_exp_stat): Remove _stat from name.
10503 * tree.h (build_vl_exp): Remove macro.
10505 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10507 * tree.c (build_decl_stat): Remove _stat from name.
10508 * tree.h (build_decl): Remove macro.
10510 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10512 * gimple.c (gimple_build_with_ops_stat): Adjust.
10513 (gimple_alloc_stat): Remove _stat from name.
10514 * gimple.h (gimple_alloc): Remove macro.
10516 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10518 * tree.c (make_tree_vec_stat): Remove _stat from name.
10519 (grow_tree_vec_stat): Likewise.
10520 * tree.h (make_tree_vec_stat): Adjust prototype.
10521 (grow_tree_vec_stat): Likewise.
10522 (make_tree_vec): Remove macro.
10523 (grow_tree_vec): Likewise.
10525 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10527 * fold-const.c (fold_build1_stat_loc): Adjust.
10528 (fold_build2_stat_loc): Likewise.
10529 (fold_build3_stat_loc): Likewise.
10530 * tree.c (build0_stat): Remove _stat from name.
10531 (build1_stat): Likewise.
10532 (build2_stat): Likewise.
10533 (build3_stat): Likewise.
10534 (build4_stat): Likewise.
10535 (build5_stat): Likewise.
10536 * tree.h (build1_loc): Remove macro, and rename _stat function
10538 (build2_loc): Likewise.
10539 (build3_loc): Likewise.
10540 (build4_loc): Likewise.
10541 (build5_loc): Likewise.
10543 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10545 * tree.c (make_int_cst_stat): Remove _stat from name.
10546 * tree.h (make_int_cst_stat): Adjust prototype.
10547 (make_int_cst): Remove macro.
10549 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10551 * tree.c (make_tre_binfo_stat): Remove _stat from name.
10552 * tree.h (make_tree_binfo_stat): Adjust prototype.
10553 (make_tree_binfo): Remove.
10555 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10557 * tree.c (copy_node_stat): Rename to copy_node.
10558 (build_distinct_type_copy): Adjust.
10559 * tree.h (copy_node_stat): Adjust prototype.
10560 (copy_node): Remove macro.
10562 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10564 * tree.c (make_node_stat): rename to make_node.
10565 (build_tree_list_stat): Adjust.
10566 (build0_stat): Likewise.
10567 (build2_stat): Likewise.
10568 (build3_stat): Likewise.
10569 (build4_stat): Likewise.
10570 (build5_stat): Likewise.
10571 (build_decl_stat): Likewise.
10572 * tree.h (make_node_stat): Adjust prototype.
10573 (make_node): remove macro.
10575 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
10577 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
10578 (PPC_FEATURE2_SCV): Likewise.
10579 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
10581 2017-07-28 Tamar Christina <tamar.christina@arm.com>
10583 * config/aarch64/aarch64.c
10584 (aarch64_internal_mov_immediate): Add new special pattern.
10585 * config/aarch64/aarch64.md (*movdi_aarch64):
10586 Add reg/32bit const mov case.
10588 2017-07-28 Tamar Christina <tamar.christina@arm.com>
10589 Richard Sandiford <richard.sandiford@linaro.org>
10591 * config/aarch64/aarch64.md (mov<mode>): Generalize.
10592 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
10593 Add integer and movi cases.
10594 (movi-split-hf-df-sf split, fp16): New.
10595 (enabled): Added TARGET_FP_F16INST.
10596 * config/aarch64/iterators.md (GPF_HF): New.
10597 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
10599 2017-07-28 Tamar Christina <tamar.christina@arm.com>
10601 * config/aarch64/aarch64.c
10602 (aarch64_simd_container_mode): Add prototype.
10603 (aarch64_expand_mov_immediate): Add HI support.
10604 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
10605 (aarch64_can_const_movi_rtx_p): New.
10606 (aarch64_preferred_reload_class):
10607 Remove restrictions of using FP registers for certain SIMD operations.
10608 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
10609 (aarch64_valid_floating_const): Add integer move validation.
10610 (aarch64_simd_imm_scalar_p): Remove.
10611 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
10612 (aarch64_legitimate_constant_p): Expand list of supported cases.
10613 * config/aarch64/aarch64-protos.h
10614 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
10615 (aarch64_reinterpret_float_as_int): New.
10616 (aarch64_simd_imm_scalar_p): Remove.
10617 * config/aarch64/constraints.md (Uvi): New.
10618 (Dd): Split into Ds and new Dd.
10619 * config/aarch64/aarch64.md (*movsi_aarch64):
10621 (*movdi_aarch64): Add SIMD mov case.
10623 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10625 * tree-predcom.c: (struct chain): Handle store-store chain in which
10626 stores for elimination only store loop invariant values.
10627 (execute_pred_commoning_chain): Ditto.
10628 (prepare_initializers_chain_store_elim): Ditto.
10629 (prepare_finalizers): Ditto.
10630 (is_inv_store_elimination_chain): New function.
10631 (initialize_root_vars_store_elim_1): New function.
10633 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10635 * tree-predcom.c: Revise general description of the pass.
10636 (enum chain_type): New enum type for store elimination.
10637 (struct chain): New field supporting store elimination.
10638 (struct component): Ditto.
10639 (dump_chain): Dump store-stores chain.
10640 (release_chain): Release resources.
10641 (split_data_refs_to_components): Compute and create component
10642 contains only stores for elimination.
10643 (get_chain_last_ref_at): New function.
10644 (make_invariant_chain): Initialization.
10645 (make_rooted_chain): Specify chain type in parameter and record it.
10646 (add_looparound_copies): Skip for store-stores chain.
10647 (determine_roots_comp): Compute type of chain and pass it to
10649 (initialize_root_vars_store_elim_2): New function.
10650 (finalize_eliminated_stores): New function.
10651 (remove_stmt): Handle store for elimination.
10652 (execute_pred_commoning_chain): Execute predictive commoning on
10653 store-store chains.
10654 (determine_unroll_factor): Skip unroll for store-stores chain.
10655 (prepare_initializers_chain_store_elim): New function.
10656 (prepare_initializers_chain): Hanlde store-store chain.
10657 (prepare_finalizers_chain, prepare_finalizers): New function.
10658 (tree_predictive_commoning_loop): Return integer value indicating
10659 if loop is unrolled or lcssa form is corrupted.
10660 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
10662 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10664 * tree-predcom.c (initialize_root): Delete.
10665 (execute_pred_commoning_chain): Initialize root vars and replace
10666 reference of non-combined chain directly, rather than call above
10669 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10671 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
10672 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
10674 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10676 * tree-predcom.c (struct chain): New field init_seq.
10677 (release_chain): Release init_seq.
10678 (prepare_initializers_chain): Record intialization stmts in above
10680 (insert_init_seqs): New function.
10681 (tree_predictive_commoning_loop): Call insert_init_seqs.
10683 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10685 * tree-predcom.c (determine_roots_comp): Skip trivial components.
10687 2017-07-28 Richard Biener <rguenther@suse.de>
10689 * match.pd: Remove superfluous :c.
10690 * genmatch.c (simplify::id): Add member.
10691 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
10693 (current_id): New global.
10694 (dt_node::parent): Move from ...
10695 (dt_operand::parent): ... here. Add for_id member.
10696 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
10697 (decision_tree::find_node): Relax order requirement when
10698 merging DT_TRUE nodes to ones inbetween the current simplify
10699 and the one we try to merge with. Add diagnostic whenever
10700 we need to enforce pattern order by not merging.
10701 (decision_tree::insert): Set current_id.
10702 (decision_tree::print_node): Dump parent node and for_id.
10703 (parser::last_id): Add member.
10704 (parser::push_simplify): Assign unique id.
10705 (parser::parser): Initialize last_id.
10707 2017-07-28 Martin Liska <mliska@suse.cz>
10710 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
10711 gimple_build_debug_bind.
10713 2017-07-28 Richard Biener <rguenther@suse.de>
10715 PR tree-optimization/81502
10716 * match.pd: Add pattern combining BIT_INSERT_EXPR with
10718 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
10720 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
10721 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
10722 for BIT_FIELD_REF args.
10723 * fold-const.c (make_bit_field_ref): Likewise.
10724 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
10726 2017-07-28 Jakub Jelinek <jakub@redhat.com>
10729 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
10730 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
10731 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
10732 Or it into SANITIZER_UNDEFINED.
10733 * ubsan.c: Include gimple-fold.h and varasm.h.
10734 (ubsan_expand_ptr_ifn): New function.
10735 (instrument_pointer_overflow): New function.
10736 (maybe_instrument_pointer_overflow): New function.
10737 (instrument_object_size): Formatting fix.
10738 (pass_ubsan::execute): Call instrument_pointer_overflow
10739 and maybe_instrument_pointer_overflow.
10740 * internal-fn.c (expand_UBSAN_PTR): New function.
10741 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
10742 * sanitizer.def (__ubsan_handle_pointer_overflow,
10743 __ubsan_handle_pointer_overflow_abort): New builtins.
10744 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
10745 * internal-fn.def (UBSAN_PTR): New internal function.
10746 * opts.c (sanitizer_opts): Add pointer-overflow.
10747 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
10748 * fold-const.c (build_range_check): Compute pointer range check in
10749 integral type if pointer arithmetics would be needed. Formatting
10752 2017-07-28 Martin Liska <mliska@suse.cz>
10755 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
10756 parameters that are of a variable-length.
10758 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
10760 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
10762 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
10763 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
10764 (CRT_CALL_STATIC_FUNCTION): Likewise.
10765 (ASM_DEFAULT_SPEC): New define.
10766 (ASM_SPEC32): Likewise.
10767 (ASM_SPEC64): Likewise.
10768 (ASM_SPEC_COMMON): Likewise.
10769 (ASM_SPEC): Likewise.
10770 (INVALID_64BIT): Likewise.
10771 (LINK_OS_DEFAULT_SPEC): Likewise.
10772 (LINK_OS_SPEC32): Likewise.
10773 (LINK_OS_SPEC64): Likewise.
10774 (POWERPC_LINUX): Likewise.
10775 (PTRDIFF_TYPE): Likewise.
10776 (RESTORE_FP_PREFIX): Likewise.
10777 (RESTORE_FP_SUFFIX): Likewise.
10778 (SAVE_FP_PREFIX): Likewise.
10779 (SAVE_FP_SUFFIX): Likewise.
10780 (SIZE_TYPE): Likewise.
10781 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
10782 (TARGET_64BIT): Likewise.
10783 (TARGET_64BIT): Likewise.
10784 (TARGET_AIX): Likewise.
10785 (WCHAR_TYPE_SIZE): Likewise.
10786 (WCHAR_TYPE): Undefine.
10787 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
10788 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
10789 (CPP_OS_RTEMS_SPEC): Delete.
10790 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
10791 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
10793 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
10795 2017-07-28 Jakub Jelinek <jakub@redhat.com>
10797 PR tree-optimization/81578
10798 * tree-parloops.c (build_new_reduction): Bail out if
10799 reduction_code isn't one of the standard OpenMP reductions.
10800 Move the details printing after that decision.
10802 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
10804 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
10805 related to reload_in_progress.
10806 (splat_input_operand): Likewise.
10807 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
10809 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
10811 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
10812 (TARGET_INSTANTIATE_DECLS): Likewise.
10813 (legitimate_indexed_address_p): Delete reload_in_progress code.
10814 (rs6000_debug_legitimate_address_p): Likewise.
10815 (rs6000_eliminate_indexed_memrefs): Likewise.
10816 (rs6000_emit_le_vsx_store): Likewise.
10817 (rs6000_emit_move_si_sf_subreg): Likewise.
10818 (rs6000_emit_move): Likewise.
10819 (register_to_reg_type): Likewise.
10820 (rs6000_pre_atomic_barrier): Likewise.
10821 (rs6000_machopic_legitimize_pic_address): Likewise.
10822 (rs6000_allocate_stack_temp): Likewise.
10823 (rs6000_address_for_fpconvert): Likewise.
10824 (rs6000_address_for_altivec): Likewise.
10825 (rs6000_secondary_memory_needed_rtx): Delete function.
10826 (rs6000_check_sdmode): Likewise.
10827 (rs6000_alloc_sdmode_stack_slot): Likewise.
10828 (rs6000_instantiate_decls): Likewise.
10829 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
10830 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
10831 Delete reload_in_progress.
10832 (*vec_reload_and_plus_<mptrsize>): Likewise.
10833 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
10834 (vsx_div_v2di): Likewise.
10835 (vsx_udiv_v2di): Likewise.
10837 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
10839 * config/rs6000/rs6000.opt (mlra): Replace with stub.
10840 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
10841 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
10842 (rs6000_debug_reg_global): Delete print of LRA status.
10843 (rs6000_option_override_internal): Delete dead LRA related code.
10844 (rs6000_lra_p): Delete function.
10845 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
10847 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
10849 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
10850 * config/riscv/rtems.h: New file.
10852 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10853 Sudakshina Das <sudi.das@arm.com>
10855 * config/aarch64/aarch64.md
10856 (define_split for and<mode>3nr_compare): Move
10857 non aarch64_logical_operand to a register.
10858 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
10859 register immediate operand to a register.
10860 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
10862 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
10864 PR middle-end/81564
10865 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
10867 2017-07-27 Richard Biener <rguenther@suse.de>
10869 PR tree-optimization/81573
10870 PR tree-optimization/81494
10871 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
10872 multi defuse cycle case.
10874 2017-07-27 Richard Biener <rguenther@suse.de>
10876 PR tree-optimization/81571
10877 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
10880 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
10882 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
10883 earlier and only if MASK_FPU is set. Adjust formatting.
10885 2017-07-27 Martin Liska <mliska@suse.cz>
10887 * opt-functions.awk: Add validation of value of Init.
10888 * optc-gen.awk: Pass new argument.
10890 2017-07-27 Martin Liska <mliska@suse.cz>
10892 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
10893 Fix wrong condition.
10895 2017-07-27 Martin Liska <mliska@suse.cz>
10897 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
10898 BBs and edges seen by autoFDO.
10900 2017-07-27 Richard Biener <rguenther@suse.de>
10902 PR tree-optimization/81502
10903 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
10904 with incompatible but same sized type.
10905 (execute_update_addresses_taken): Likewise.
10907 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
10909 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
10910 flag_tree_loop_vectorize rather than flag_tree_vectorize.
10912 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10915 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
10916 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
10917 Change s_operand to memory_operand.
10919 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
10921 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
10922 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
10923 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
10924 Emit instructions rather than returning an expression. Handle TFmode
10925 and KFmode by casting to TImode.
10926 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
10927 (rs6000_emit_le_vsx_store): Likewise.
10928 * config/rs6000/vsx.md (VSX_TI): New iterator.
10929 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
10930 (*vsx_le_undo_permute_<mode>): Likewise.
10931 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
10932 emit the split sequence.
10933 (*vsx_le_perm_store_<mode>): Likewise.
10935 2017-07-27 Jakub Jelinek <jakub@redhat.com>
10937 PR tree-optimization/81555
10938 PR tree-optimization/81556
10939 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
10940 if true, force CHANGED for the recursive invocation.
10941 (reassociate_bb): Remember original length of ops array, pass
10942 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
10944 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
10945 attributes for noipa attribute. For naked attribute use
10946 lookup_attribute first before lookup_attribute_spec.
10947 * final.c (rest_of_handle_final): Disable IPA RA for functions with
10949 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
10950 for functions with noipa attribute.
10951 (cgraph_externally_visible_p): Return true for functions with noipa
10953 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
10954 for functions with noipa attribute.
10955 * doc/extend.texi: Document noipa function attribute.
10956 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
10957 also for functions with noipa attribute.
10958 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
10960 2017-07-26 Andrew Pinski <apinski@cavium.com>
10962 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
10963 vec_unalign_load_cost and vec_unalign_store_cost.
10965 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10967 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
10968 -mvsx-small-integer option.
10969 (ISA_3_0_MASKS_IEEE): Likewise.
10970 (OTHER_VSX_VECTOR_MASKS): Likewise.
10971 (POWERPC_MASKS): Likewise.
10972 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
10973 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
10974 code, only testing for DImode being allowed in non-VSX floating
10976 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
10977 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
10979 (rs6000_option_override_internal): Delete -mvsx-small-integer.
10980 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
10981 TARGET_P8_VECTOR test.
10982 (rs6000_secondary_reload_simple_move): Likewise.
10983 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
10984 since TARGET_P9_VECTOR was already tested.
10985 (rs6000_opt_masks): Remove -mvsx-small-integer.
10986 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
10987 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
10989 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
10990 test for TARGET_VEXTRACTUB was used, and that uses
10992 (p9 extract splitter): Likewise.
10993 (vsx_extract_<mode>_di_p9): Likewise.
10994 (vsx_extract_<mode>_store_p9): Likewise.
10995 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
10996 for TARGET_P9_VECTOR was used. Delete code that is now dead with
10997 the elimination of TARGET_VSX_SMALL_INTEGER.
10998 (vsx_extract_<mode>_p8): Likewise.
10999 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
11000 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
11001 (vsx_set_<mode>_p9): Likewise.
11002 (vsx_set_v4sf_p9): Likewise.
11003 (vsx_set_v4sf_p9_zero): Likewise.
11004 (vsx_insert_extract_v4sf_p9): Likewise.
11005 (vsx_insert_extract_v4sf_p9_2): Likewise.
11006 * config/rs6000/rs6000.md (sign extend splitter): Change
11007 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
11008 (floatsi<mode>2_lfiwax_mem): Likewise.
11009 (floatunssi<mode>2_lfiwzx_mem): Likewise.
11010 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
11011 since a test for TARGET_P9_VECTOR was used.
11012 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
11013 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
11014 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
11015 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
11016 TARGET_P8_VECTOR test.
11017 (fix_trunc<mode>si2_stfiwx): Likewise.
11018 (fix_trunc<mode>si2_internal): Likewise.
11019 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
11020 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
11022 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
11023 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
11024 TARGET_P8_VECTOR test.
11025 (fixuns_trunc<mode>si2_stfiwx): Likewise.
11026 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
11027 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
11029 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
11030 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
11031 since a test for TARGET_P9_VECTOR was used.
11032 (splitter for loading small constants): Likewise.
11034 2017-07-26 Andrew Pinski <apinski@cavium.com>
11036 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
11039 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
11042 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
11043 (fp_valid_at): Likewise.
11045 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
11047 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
11048 (qdf24xx_addrcost_table): Likewise.
11049 (cortexa57_tunings): Update to use generic_branch_cost.
11050 (cortexa72_tunings): Likewise.
11051 (cortexa73_tunings): Likewise.
11052 (qdf24xx_tunings): Likewise.
11054 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
11056 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
11057 (thunderx2t99_branch_cost): Likewise.
11058 (cortexa35_tunings): Update to use generic_branch_cost.
11059 (cortexa53_tunings): Likewise.
11060 (cortexa57_tunings): Likewise.
11061 (cortexa72_tunings): Likewise.
11062 (cortexa73_tunings): Likewise.
11063 (thunderx2t99_tunings): Likewise.
11065 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
11067 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
11068 (sparc_option_override): Honour MASK_FSMULD.
11069 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
11070 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
11071 * config/sparc/sparc.opt (mfsmuld): New option.
11072 * doc/invoke.texi (mfsmuld): Document option.
11074 2017-07-26 Marek Polacek <polacek@redhat.com>
11076 PR middle-end/70992
11077 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
11079 2017-07-26 Richard Biener <rguenther@suse.de>
11081 * gimple-match-head.c (do_valueize): Return OP if valueize
11083 (get_def): New helper to get at the def stmt of a SSA name
11084 if valueize allows.
11085 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
11086 do_valueize to get at the def stmt.
11087 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
11089 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
11091 PR middle-end/46932
11092 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
11094 2017-07-26 Martin Liska <mliska@suse.cz>
11097 * function.c (expand_function_start): Make expansion of
11098 nonlocal_goto_save_area after parm_birth_insn.
11100 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
11102 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
11103 from all CPU target flags enable members.
11105 2017-07-26 Richard Biener <rguenther@suse.de>
11107 * genmatch.c (dt_simplify::gen): Make iterator vars const.
11108 (decision_tree::gen): Make 'type' const.
11109 (write_predicate): Likewise.
11111 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
11113 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
11114 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
11115 (rs6000_option_override_internal): Likewise.
11116 (rs6000_expand_vector_set): Likewise.
11117 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
11118 (TARGET_UPPER_REGS_SF): Likewise.
11119 (TARGET_UPPER_REGS_DI): Likewise.
11120 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
11121 (TARGET_DIRECT_MOVE_64BIT): Likewise.
11122 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
11123 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
11124 (Splitters for DI constants in Altivec registers): Likewise.
11125 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
11126 (vsx_set_v4sf_p9): Likewise.
11127 (vsx_set_v4sf_p9_zero): Likewise.
11128 (vsx_insert_extract_v4sf_p9): Likewise.
11129 (vsx_insert_extract_v4sf_p9_2): Likewise.
11131 2017-07-25 Carl Love <cel@us.ibm.com>
11133 * doc/extend.texi: Update the built-in documentation file for the
11134 existing built-in functions
11135 vector signed char vec_cnttz (vector signed char);
11136 vector unsigned char vec_cnttz (vector unsigned char);
11137 vector signed short vec_cnttz (vector signed short);
11138 vector unsigned short vec_cnttz (vector unsigned short);
11139 vector signed int vec_cnttz (vector signed int);
11140 vector unsigned int vec_cnttz (vector unsigned int);
11141 vector signed long long vec_cnttz (vector signed long long);
11142 vector unsigned long long vec_cnttz (vector unsigned long long);
11144 2017-07-25 Andrew Pinski <apinski@cavium.com>
11146 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
11147 accesses where the use is for the first operand of a BIT_INSERT.
11149 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
11152 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
11153 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
11155 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
11157 * config/i386/gstabs.h: Delete.
11158 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
11160 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
11162 * config/i386/i386.c (ix86_decompose_address): Do not check for
11163 register RTX when looking at index_reg or base_reg.
11164 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
11166 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
11168 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
11169 to update EH info here.
11171 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
11173 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
11175 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
11177 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
11179 2017-07-25 Torsten Duwe <duwe@suse.de>
11181 * common.opt: Introduce -fpatchable-function-entry
11182 command line option, and its variables function_entry_patch_area_size
11183 and function_entry_patch_area_start.
11184 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
11185 including a two-value parser.
11186 * target.def (print_patchable_function_entry): New target hook.
11187 * targhooks.h (default_print_patchable_function_entry): New function.
11188 * targhooks.c (default_print_patchable_function_entry): Likewise.
11189 * toplev.c (process_options): Switch off IPA-RA if
11190 patchable function entries are being generated.
11191 * varasm.c (assemble_start_function): Look at the
11192 patchable-function-entry command line switch and current
11193 function attributes and maybe generate NOP instructions by
11194 calling the print_patchable_function_entry hook.
11195 * doc/extend.texi: Document patchable_function_entry attribute.
11196 * doc/invoke.texi: Document -fpatchable_function_entry
11197 command line option.
11198 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
11200 * doc/tm.texi: Re-generate.
11202 2017-07-25 Jakub Jelinek <jakub@redhat.com>
11205 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
11206 TARGET_AVX512DQ rather than TARGET_AVX512BW.
11208 2017-07-25 Tamar Christina <tamar.christina@arm.com>
11210 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
11212 2017-07-25 Richard Biener <rguenther@suse.de>
11214 PR tree-optimization/81455
11215 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
11216 not walk in cycles when looking for guards.
11218 2017-07-25 Richard Biener <rguenther@suse.de>
11220 PR tree-optimization/81529
11221 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
11222 when optimizing backedge uses.
11224 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
11226 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
11228 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
11229 to dl_section_ref. On AIX, append an expression to subtract
11230 the size of the section length to dl_section_ref.
11232 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
11234 * configure.ac: If any of the config.* scripts fail, exit 1.
11235 * configure: Regenerate.
11237 2017-07-25 Richard Biener <rguenther@suse.de>
11239 PR middle-end/81546
11240 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
11241 of immediate uses, be more verbose on errors.
11243 2017-07-25 Richard Biener <rguenther@suse.de>
11245 PR tree-optimization/81510
11246 * tree-vect-loop.c (vect_is_simple_reduction): When the
11247 reduction stmt is not inside the loop bail out.
11249 2017-07-25 Richard Biener <rguenther@suse.de>
11251 PR tree-optimization/81303
11252 * tree-vect-loop-manip.c (vect_loop_versioning): Build
11253 profitability check against LOOP_VINFO_NITERSM1.
11255 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
11257 * domwalk.c (cmp_bb_postorder): Simplify.
11258 (sort_bbs_postorder): New function. Use it...
11259 (dom_walker::walk): ...here to optimize common cases.
11261 2017-07-25 Martin Liska <mliska@suse.cz>
11264 * ipa-visibility.c (function_and_variable_visibility): Make the
11265 redirection just on target that supports aliasing.
11266 Fix GNU coding style.
11268 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
11271 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
11273 * config/arm/rtems.h: Likewise.
11274 * config/bfin/rtems.h: Likewise.
11275 * config/i386/rtemself.h: Likewise.
11276 * config/lm32/rtems.h: Likewise.
11277 * config/m32c/rtems.h: Likewise.
11278 * config/m68k/rtemself.h: Likewise.
11279 * config/microblaze/rtems.h: Likewise.
11280 * config/mips/rtems.h: Likewise.
11281 * config/moxie/rtems.h: Likewise.
11282 * config/nios2/rtems.h: Likewise.
11283 * config/powerpcspe/rtems.h: Likewise.
11284 * config/rs6000/rtems.h: Likewise.
11285 * config/rtems.h: Likewise.
11286 * config/sh/rtems.h: Likewise.
11287 * config/sh/rtemself.h: Likewise.
11288 * config/sparc/rtemself.h: Likewise.
11290 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
11293 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
11294 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
11295 * tree-ssa-structalias.c (alias_get_name): Same.
11297 2017-07-25 Bin Cheng <bin.cheng@arm.com>
11300 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
11301 instructions if no du chain is found.
11303 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
11305 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
11307 2017-07-25 Richard Biener <rguenther@suse.de>
11309 PR middle-end/81505
11310 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
11313 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
11315 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
11316 upper-regs options.
11317 (ISA_2_7_MASKS_SERVER): Likewise.
11318 (ISA_3_0_MASKS_IEEE): Likewise.
11319 (OTHER_P8_VECTOR_MASKS): Likewise.
11320 (OTHER_VSX_VECTOR_MASKS): Likewise.
11321 (POWERPC_MASKS): Likewise.
11322 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
11323 duplicate list of options.
11324 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
11325 explicit -mupper-regs options.
11326 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
11327 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
11328 alias for -mupper-regs-df.
11329 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
11330 (rs6000_init_hard_regno_mode_ok): Likewise.
11331 (rs6000_option_override_internal): Likewise.
11332 (rs6000_opt_masks): Likewise.
11333 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
11334 options in terms of whether -mvsx or -mpower8-vector was used.
11335 (TARGET_UPPER_REGS_DI): Likewise.
11336 (TARGET_UPPER_REGS_SF): Likewise.
11337 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
11338 -mupper-regs-* options.
11340 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
11342 * passes.c (emergency_dump_function): Print some empty lines and a
11343 header before the RTL dump.
11345 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
11347 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
11349 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
11352 * config/aarch64/aarch64.c (aarch64_classify_symbol):
11353 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
11355 2017-07-24 Carl Love <cel@us.ibm.com>
11357 * config/rs6000/rs6000-c.c: Add support for built-in functions
11358 vector float vec_extract_fp32_from_shorth (vector unsigned short);
11359 vector float vec_extract_fp32_from_shortl (vector unsigned short);
11360 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
11361 vec_extract_fp_from_shortl): Add defines for the two builtins.
11362 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
11363 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
11365 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
11366 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
11367 * doc/extend.texi: Update the built-in documentation file for the
11368 new built-in function.
11370 2017-07-24 Jakub Jelinek <jakub@redhat.com>
11373 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
11375 * doc/generic.texi: Likewise.
11376 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
11377 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
11379 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
11381 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
11382 (aarch64_mls_elt_merge<mode>): Likewise.
11384 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
11386 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
11387 having __cxa_atexit.
11389 2017-07-23 Michael Collison <michael.collison@arm.com>
11391 * config/arm/arm.c (arm_option_override): Deprecate
11392 use of -mstructure-size-boundary.
11393 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
11394 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
11396 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11399 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
11400 Reduce cost estimate for direct moves.
11402 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
11405 * config/i386/i386.c (ix86_option_override_internal): Disable
11406 BMI, BMI2 and TBM instructions for -m16.
11408 2017-07-21 Carl Love <cel@us.ibm.com>
11410 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11411 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11412 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11413 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11414 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11415 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
11416 VMULOSW): New enum "unspec" values.
11417 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
11418 altivec_vmulosw): New patterns.
11419 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11420 VMULOSW): Add definitions.
11422 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
11424 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
11425 (qdf24xx): Likewise.
11426 * config/aarch64/aarch64-options-extensions.def (rdma); New.
11427 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
11428 (AARCH64_FL_V8_1): Renumber.
11429 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
11430 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
11431 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
11432 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
11433 rdma to feature modifiers list.
11435 2017-07-21 Yury Gribov <tetra2005@gmail.com>
11437 PR middle-end/56727
11438 * ipa-visibility (function_and_variable_visibility): Convert
11439 recursive PLT call to direct call if appropriate.
11441 2017-07-21 Andrew Pinski <apinski@cavium.com>
11443 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
11444 operand 1 to see if the types precision matches.
11445 * fold-const.c (operand_equal_p): Likewise.
11447 2017-07-21 Richard Biener <rguenther@suse.de>
11449 PR tree-optimization/81303
11450 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
11451 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
11452 (vect_peeling_hash_get_lowest_cost): Adjust.
11453 (vect_enhance_data_refs_alignment): Likewise. Use
11454 vect_get_peeling_costs_all_drs to compute the penalty for no
11455 peeling to match up costs.
11457 2017-07-21 Richard Biener <rguenther@suse.de>
11459 PR tree-optimization/81500
11460 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
11461 we didn't identify a reduction path.
11463 2017-07-21 Tom de Vries <tom@codesourcery.com>
11464 Cesar Philippidis <cesar@codesourcery.com>
11466 PR gcov-profile/81442
11467 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
11470 2017-07-21 Tom de Vries <tom@codesourcery.com>
11473 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
11475 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
11476 nvptx_override_options_after_change.
11478 2017-07-21 Ulrich Drepper <drepper@redhat.com>
11480 * dwarf2out.c (output_file_names): Avoid double testing for
11481 dwarf_version >= 5.
11483 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
11485 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
11487 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
11489 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
11491 (try_crossjump_to_edge): Do not punt on partitioned functions.
11493 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
11495 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
11496 Put all BBs reachable only via paths crossing cold region to cold
11498 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
11500 2016-07-21 Richard Biener <rguenther@suse.de>
11502 PR tree-optimization/81303
11503 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
11504 into account prologue and epilogue iterations when raising
11505 min_profitable_iters to sth at least covering one vector iteration.
11507 2017-07-21 Tamar Christina <tamar.christina@arm.com>
11509 * config/arm/arm.c (arm_test_cpu_arch_dat):
11512 2017-07-20 Nathan Sidwell <nathan@acm.org>
11514 Remove TYPE_METHODS.
11515 * tree.h (TYPE_METHODS): Delete.
11516 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
11517 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
11518 (dbxout_type_methods): Scan TYPE_FIELDS.
11519 (dbxout_type): Don't check TYPE_METHODS here.
11520 * function.c (use_register_for_decl): Always ignore register for
11521 class types when not optimizing.
11522 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
11523 * tree.c (free_lang_data_in_type): Stitch out member functions and
11524 templates from TYPE_FIELDS.
11525 (build_distinct_type_copy, verify_type_variant,
11526 verify_type): Member fns are on TYPE_FIELDS.
11527 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
11528 * tree-pretty-print.c (dump_generic_node): Likewise.
11530 2017-07-20 Jakub Jelinek <jakub@redhat.com>
11533 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
11534 V2TImode and V4TImode.
11535 (ix86_expand_vector_extract): Likewise.
11536 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
11537 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
11538 (ssescalarmode): Handle V4TImode and V2TImode.
11539 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
11540 (*vec_extractv2ti, *vec_extractv4ti): New insns.
11541 (VEXTRACTI128_MODE): New mode iterator.
11542 (splitter for *vec_extractv?ti first element): New.
11543 (VEC_INIT_MODE): New mode iterator.
11544 (vec_init<mode>): Consolidate 3 expanders into one using
11545 VEC_INIT_MODE mode iterator.
11547 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
11549 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
11550 non_spilled_static_chain_regno_p.
11552 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
11554 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
11556 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
11558 * bb-reorder.c (connect_traces): Allow copying of blocks within
11561 2017-07-20 Richard Biener <rguenther@suse.de>
11563 * gimple.h (gimple_phi_result): Add gphi * overload.
11564 (gimple_phi_result_ptr): Likewise.
11565 (gimple_phi_arg): Likewise. Adjust index assert to only
11566 allow actual argument accesses rather than all slots available
11568 (gimple_phi_arg_def): Add gphi * overload.
11569 * tree-phinodes.c (make_phi_node): Initialize only actual
11571 (resize_phi_node): Clear memory not covered by old node,
11572 do not initialize excess argument slots.
11573 (reserve_phi_args_for_new_edge): Initialize new argument slot
11576 2017-07-20 Bin Cheng <bin.cheng@arm.com>
11578 PR tree-optimization/81388
11580 2016-07-21 Bin Cheng <bin.cheng@arm.com>
11582 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
11583 by removing computation of may_be_zero.
11585 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
11586 Tom de Vries <tom@codesourcery.com>
11588 PR middle-end/81030
11589 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
11590 when gimple level profile disagrees with what RTL expander did.
11592 2017-07-20 Richard Biener <rguenther@suse.de>
11594 PR tree-optimization/61171
11595 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
11596 (vect_analyze_stmt): Add slp instance parameter.
11597 (vectorizable_reduction): Likewise.
11598 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
11599 (vect_is_simple_reduction): Deal with chains not detected
11600 as SLP reduction chain, specifically not properly associated
11601 chains containing a mix of plus/minus.
11602 (get_reduction_op): Remove.
11603 (get_initial_defs_for_reduction): Simplify, pass in whether
11604 this is a reduction chain, pass in the SLP node for the PHIs.
11605 (vect_create_epilog_for_reduction): Get the SLP instance as
11607 (vectorizable_reduction): Get the SLP instance as arg.
11608 During analysis remember the SLP node with the PHIs in the
11609 instance. Simplify getting at the vectorized reduction PHIs.
11610 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
11611 through SLP instance.
11612 (vect_slp_analyze_operations): Likewise.
11613 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
11614 (vect_transform_stmt): Likewise.
11616 2017-07-20 Tom de Vries <tom@codesourcery.com>
11618 PR tree-optimization/81489
11619 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
11620 read of phi arg location to before loop that modifies phi.
11622 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11624 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
11627 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
11629 PR middle-end/81331
11630 * except.c (execute): Fix ordering issue.
11632 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
11634 PR rtl-optimization/81423
11635 * combine.c (make_compound_operation_int): Don't try to optimize
11636 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
11638 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
11640 PR rtl-optimization/81423
11641 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
11642 with a constant that is -1 in the truncated to mode.
11644 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
11646 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
11647 (determine_unlikely_bbs): ... here.
11648 * predict.h (propagate_unlikely_bbs_forward): Declare.
11649 * cfgexpand.c (pass_expand::execute): Use it.
11650 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
11652 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
11653 propagate_unlikely_bbs_forward.
11655 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
11657 PR middle-end/81331
11658 * except.c (maybe_add_nop_after_section_switch): New function.
11661 2017-07-19 Tom de Vries <tom@codesourcery.com>
11663 * gimple.h (gimple_phi_set_arg): Make assert more strict.
11665 2017-07-19 Tom de Vries <tom@codesourcery.com>
11667 * gimple.h (gimple_phi_arg): Make assert more strict.
11669 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
11671 * config.gcc (powerpc*-*-*): Add mmintrin.h.
11672 * config/rs6000/mmintrin.h: New file.
11673 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
11675 2017-07-19 Jakub Jelinek <jakub@redhat.com>
11677 PR tree-optimization/81346
11678 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
11680 2017-07-19 Tom de Vries <tom@codesourcery.com>
11682 * config/nvptx/nvptx.md (VECIM): Add V2DI.
11684 2017-07-19 Tom de Vries <tom@codesourcery.com>
11686 * config/nvptx/nvptx-modes.def: Add V2DImode.
11687 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
11688 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
11689 (nvptx_output_mov_insn): Handle lack of mov.b128.
11690 (nvptx_print_operand): Handle 'H' and 'L' codes.
11691 (nvptx_vector_mode_supported): Allow V2DImode.
11692 (nvptx_preferred_simd_mode): New function.
11693 (nvptx_data_alignment): New function.
11694 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
11695 nvptx_preferred_simd_mode.
11696 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
11698 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
11700 2017-07-19 Tom de Vries <tom@codesourcery.com>
11702 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
11703 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
11704 (nvptx_vector_mode_supported): New function. Allow V2SImode.
11705 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
11706 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
11707 (mov<VECIM>_insn): New define_insn.
11708 (define_expand "mov<VECIM>): New define_expand.
11710 2017-07-19 Tom de Vries <tom@codesourcery.com>
11712 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
11714 2017-07-19 Jakub Jelinek <jakub@redhat.com>
11716 PR tree-optimization/81346
11717 * fold-const.h (fold_div_compare, range_check_type): Declare.
11718 * fold-const.c (range_check_type): New function.
11719 (build_range_check): Use range_check_type.
11720 (fold_div_compare): No longer static, rewritten into
11721 a match.pd helper function.
11722 (fold_comparison): Don't call fold_div_compare here.
11723 * match.pd (X / C1 op C2): New optimization using fold_div_compare
11724 as helper function.
11726 2017-07-19 Nathan Sidwell <nathan@acm.org>
11728 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
11729 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
11730 * tree.c (find_decls_types_r, verify_type): Use
11731 TYPE_{MIN,MAX}_VALUE_RAW.
11732 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
11733 (hash_tree): Likewise.
11734 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
11736 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
11739 2017-07-18 Tom de Vries <tom@codesourcery.com>
11741 PR middle-end/81464
11742 * omp-expand.c (expand_omp_for_static_chunk): Handle
11743 equal-argument loop exit phi.
11745 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
11748 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
11749 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
11750 operand 2 predicate.
11751 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
11752 operand 2 predicate.
11753 (ror,rol -> rorx splitters): Use const_int_operand as
11754 operand 2 predicate.
11756 2017-06-18 Richard Biener <rguenther@suse.de>
11758 PR tree-optimization/81410
11759 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
11760 the gap in the ! slp_perm SLP case after each group.
11762 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
11764 PR middle-end/81463
11765 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
11768 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
11770 PR middle-end/81462
11771 * predict.c (set_even_probabilities): Cleanup; do not affect
11772 probabilities that are already known.
11773 (combine_predictions_for_bb): Call even when count is set.
11775 2017-07-18 Nathan Sidwell <nathan@acm.org>
11777 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
11780 2017-07-18 Bin Cheng <bin.cheng@arm.com>
11783 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
11784 optimization for loop niter analysis.
11786 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
11789 * config/avr/avr.c (avr_optimize_casesi): Don't use
11790 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
11792 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11794 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
11795 body_cost_vec from _vect_peel_extended_info.
11796 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
11797 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
11800 2017-07-18 Bin Cheng <bin.cheng@arm.com>
11802 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
11804 2017-07-18 Richard Biener <rguenther@suse.de>
11806 PR tree-optimization/80620
11807 PR tree-optimization/81403
11808 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
11809 info when re-using a VN table entry.
11811 2017-07-18 Richard Biener <rguenther@suse.de>
11813 PR tree-optimization/81418
11814 * tree-vect-loop.c (vectorizable_reduction): Properly compute
11815 vectype_in. Verify that with lane-reducing reduction operations
11816 we have a single def-use cycle.
11818 2017-07-17 Carl Love <cel@us.ibm.com>
11820 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
11822 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11823 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11824 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11825 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11826 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11827 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
11828 VMULOSW): New enum "unspec" values.
11829 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
11830 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
11831 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
11832 altivec_vmulosw): New patterns.
11833 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11834 VMULOSW): Add definitions.
11835 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
11837 * config/alpha/alpha.c: Include predict.h.
11839 2017-07-17 Yury Gribov <tetra2005@gmail.com>
11841 * tree-vrp.c (compare_assert_loc): Fix comparison function
11842 to return predictable results.
11844 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11846 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
11848 (subdi3): Likewise.
11849 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
11850 * doc/invoke.texi (mexpand-adddi): Update text.
11852 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11854 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
11855 that also clobbers the CC register. The old expand code is moved
11857 (*arc_clzsi2): ... here.
11858 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
11859 the CC register. The old expand code is moved to ...
11860 (arc_ctzsi2): ... here.
11862 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11864 * config/arc/arc.opt (mindexed-loads): Use initial value
11865 TARGET_INDEXED_LOADS_DEFAULT.
11866 (mauto-modify-reg): Use initial value
11867 TARGET_AUTO_MODIFY_REG_DEFAULT.
11868 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
11869 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
11870 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
11871 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
11873 2017-07-17 Martin Liska <mliska@suse.cz>
11876 * opts.c (finish_options): Do not allow -fgnu-tm
11877 w/ -fsanitize={kernel-,}address. Say sorry.
11879 2017-07-17 Bin Cheng <bin.cheng@arm.com>
11882 * tree-loop-distribution.c (classify_partition): Only assert on
11883 numer of iterations.
11884 (merge_dep_scc_partitions): Delete prameter. Update function call.
11885 (distribute_loop): Remove code handling loop with unknown niters.
11886 (pass_loop_distribution::execute): Skip loop with unknown niters.
11888 2017-07-17 Bin Cheng <bin.cheng@arm.com>
11891 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
11892 function sort_partitions_by_post_order.
11894 2017-07-17 Bin Cheng <bin.cheng@arm.com>
11896 PR tree-optimization/81374
11897 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
11898 the max index of basic blocks, rather than number of basic blocks.
11900 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11902 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
11904 (arc_legitimate_pic_operand_p): Likewise.
11905 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
11907 (arc_needs_pcl_p): Likewise.
11908 (arc_legitimate_pc_offset_p): Likewise.
11909 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
11910 function is also used in constrains.md.
11911 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
11912 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
11913 PLUS. Only return true/false in known cases, otherwise assert.
11914 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
11915 is already called in arc_legitimate_constant_p.
11916 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
11918 (LEGITIMATE_PIC_OPERAND_P): Use
11919 arc_raw_symbolic_reference_mentioned_p function.
11920 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
11925 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11926 Andrew Burgess <andrew.burgess@embecosm.com>
11928 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
11929 (arc_return_address_register): New function.
11930 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
11931 (arc_handle_fndecl_attribute): Add naked attribute.
11932 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
11933 (TARGET_WARN_FUNC_RETURN): Likewise.
11934 (arc_allocate_stack_slots_for_args): New function.
11935 (arc_warn_func_return): Likewise.
11936 (machine_function): Change type fn_type.
11937 (arc_compute_function_type): Consider new naked function type,
11938 change function return type.
11939 (arc_must_save_register): Adapt to handle new
11940 arc_compute_function_type's return type.
11941 (arc_expand_prologue): Likewise.
11942 (arc_expand_epilogue): Likewise.
11943 (arc_return_address_regs): Delete.
11944 (arc_return_address_register): New function.
11945 (arc_epilogue_uses): Use above function.
11946 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
11947 (arc_function_type): Change encoding, add naked type.
11948 (ARC_INTERRUPT_P): Change to handle the new encoding.
11949 (ARC_FAST_INTERRUPT_P): Likewise.
11950 (ARC_NORMAL_P): Define.
11951 (ARC_NAKED_P): Likewise.
11952 (arc_compute_function_type): Delete prototype.
11953 * config/arc/arc.md (in_ret_delay_slot): Use
11954 arc_return_address_register function.
11955 (simple_return): Likewise.
11956 (p_return_i): Likewise.
11958 2017-07-17 Jakub Jelinek <jakub@redhat.com>
11960 PR tree-optimization/81428
11961 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
11962 can't be built for those types.
11964 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
11966 Remove stuff dead since r239246.
11968 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
11969 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
11970 (avr_inform_devices): Remove dead stuff.
11972 2017-07-17 Tamar Christina <tamar.christina@arm.com>
11974 * config/arm/arm_neon.h: Fix softp typo.
11976 2017-07-17 Jakub Jelinek <jakub@redhat.com>
11978 PR tree-optimization/81365
11979 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
11980 aggregate moves onto bb predecessor edges, make sure there are no
11981 loads that could alias the lhs in between the start of bb and the
11984 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
11987 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
11988 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
11989 [LSHIFTRT, outer_code = TRUNCATE]: Same.
11991 2017-07-17 Jakub Jelinek <jakub@redhat.com>
11993 PR tree-optimization/81396
11994 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
11995 (init_symbolic_number): Initialize it to 1.
11996 (perform_symbolic_merge): Add n_ops from both operands into the new
11998 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
11999 without base_addr as useless if they need more than one operation.
12000 (bswap_replace): Handle !bswap case for NULL base_addr.
12002 2017-07-17 Tom de Vries <tom@codesourcery.com>
12005 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
12008 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
12010 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
12011 conditional builtin define __FIX_LEON3FT_B2BST.
12013 2017-07-17 Daniel Cederman <cederman@gaisler.com>
12015 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
12016 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
12019 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
12021 PR rtl-optimization/81424
12022 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
12023 to remove potential trapping from operands if -fnon-call-exceptions.
12025 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12027 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
12028 profile_proability for scalling.
12029 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
12031 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12033 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
12035 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12037 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
12038 fixpoint arithmetics.
12040 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12042 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
12043 fixpoint arithmetics.
12045 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12047 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
12048 fixpoint arithmetics.
12050 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12052 * profile-count.h (profile_probability::from_reg_br_prob_note,
12053 profile_probability::to_reg_br_prob_note): New functions.
12054 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
12055 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
12056 * predict.c (probability_reliable_p): Update.
12057 (edge_probability_reliable_p): Update.
12058 (br_prob_note_reliable_p): Update.
12059 (invert_br_probabilities): Update.
12060 (add_reg_br_prob_note): New function.
12061 (combine_predictions_for_insn): Update.
12062 * asan.c (asan_clear_shadow): Update.
12063 * cfgbuild.c (compute_outgoing_frequencies): Update.
12064 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
12065 (update_br_prob_note): Update.
12066 (rtl_verify_edges): Update.
12067 (purge_dead_edges): Update.
12068 (fixup_reorder_chain): Update.
12069 * emit-rtl.c (try_split): Update.
12070 * ifcvt.c (cond_exec_process_insns): Update.
12071 (cond_exec_process_if_block): Update.
12072 (dead_or_predicable): Update.
12073 * internal-fn.c (expand_addsub_overflow): Update.
12074 (expand_neg_overflow): Update.
12075 (expand_mul_overflow): Update.
12076 * loop-doloop.c (doloop_modify): Update.
12077 * loop-unroll.c (compare_and_jump_seq): Update.
12078 * optabs.c (emit_cmp_and_jump_insn_1): Update.
12079 * predict.h: Update.
12080 * reorg.c (mostly_true_jump): Update.
12082 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
12083 * config/alpha/alpha.c (emit_unlikely_jump): Update.
12084 * config/arc/arc.c: (emit_unlikely_jump): Update.
12085 * config/arm/arm.c: (emit_unlikely_jump): Update.
12086 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
12087 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
12088 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
12089 (ix86_print_operand): Update.
12090 (ix86_split_fp_branch): Update.
12091 (predict_jump): Update.
12092 * config/ia64/ia64.c (ia64_print_operand): Update.
12093 * config/mmix/mmix.c (mmix_print_operand): Update.
12094 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
12095 (rs6000_expand_split_stack_prologue): Update.
12096 * config/rs6000/rs6000.c: Update.
12097 * config/s390/s390.c (s390_expand_vec_strlen): Update.
12098 (s390_expand_vec_movstr): Update.
12099 (s390_expand_cs_tdsi): Update.
12100 (s390_expand_split_stack_prologue): Update.
12101 * config/sh/sh.c (sh_print_operand): Update.
12102 (expand_cbranchsi4): Update.
12103 (expand_cbranchdi4): Update.
12104 * config/sparc/sparc.c (output_v9branch): Update.
12105 * config/spu/spu.c (get_branch_target): Update.
12106 (ea_load_store_inline): Update.
12107 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
12108 * config/tilepro/tilepro.c: Update.
12110 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
12112 * gimplify.c (mostly_copy_tree_r): Revert latest change.
12113 (gimplify_save_expr): Likewise.
12115 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
12117 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
12119 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
12121 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
12123 * timevar.def (TV_IPA_FNSUMMARY): Define.
12125 2017-07-16 Daniel Cederman <cederman@gaisler.com>
12127 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
12128 to back store errata sensitive sequence from being generated.
12129 (sqrtdf2_fix): Likewise.
12131 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
12133 * tree-ssa-threadupdate.c (compute_path_counts,
12134 update_joiner_offpath_counts): Use profile_probability.
12136 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
12139 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
12141 * config/arm/arm-c.c (arm_cpu_builtins): Define
12142 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
12144 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
12146 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12147 array entries to represent __ieee128 versions of the
12148 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
12149 scalar_extract_sig, and scalar_insert_exp built-in functions.
12150 (altivec_resolve_overloaded_builtin): Add special case handling
12151 for the __builtin_scalar_insert_exp function, as represented by
12152 the P9V_BUILTIN_VEC_VSIEDP constant.
12153 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
12154 exponent support for __ieee128 argument.
12155 (VSESQP): Add scalar extract signature support for __ieee128
12157 (VSTDCNQP): Add scalar test negative support for __ieee128
12159 (VSIEQP): Add scalar insert exponent support for __int128 argument
12160 with __ieee128 result.
12161 (VSIEQPF): Add scalar insert exponent support for __ieee128
12162 argument with __ieee128 result.
12163 (VSTDCQP): Add scalar test data class support for __ieee128
12165 (VSTDCNQP): Add overload support for scalar test negative with
12166 __ieee128 argument.
12167 (VSTDCQP): Add overload support for scalar test data class
12168 __ieee128 argument.
12169 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
12170 UNSPEC_VSX_SXSIGDP.
12171 (UNSPEC_VSX_SIEXPQP): New constant.
12172 (xsxexpqp): New insn for VSX scalar extract exponent quad
12174 (xsxsigqp): New insn for VSX scalar extract significand quad
12176 (xsiexpqpf): New insn for VSX scalar insert exponent quad
12177 precision with floating point argument.
12178 (xststdcqp): New expand for VSX scalar test data class quad
12180 (xststdcnegqp): New expand for VSX scalar test negative quad
12182 (xststdcqp): New insn to match expansions for VSX scalar test data
12183 class quad precision and VSX scalar test negative quad precision.
12184 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
12185 special case operand checking to enforce that second operand of
12186 VSX scalar test data class with quad precision argument is a 7-bit
12188 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
12189 prototypes and descriptions of __ieee128 versions of
12190 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
12191 scalar_test_data_class, and scalar_test_neg built-in functions.
12193 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12195 PR tree-optimization/81162
12196 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
12197 replace a negate with an add.
12199 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
12201 * doc/invoke.texi (arm/-mcpu): Document +crypto.
12203 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
12205 * config/arm/arm-c.c (arm_cpu_builtins): Define
12206 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
12208 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
12210 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
12211 (armv8-r): Set ARM Cortex-R52 as default CPU.
12212 * config/arm/arm-tables.opt: Regenerate.
12213 * config/arm/arm-tune.md: Regenerate.
12214 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
12216 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
12217 extension for -mcpu=cortex-r52.
12219 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
12221 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
12222 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
12223 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
12224 (fp-armv8): Define it as FP_ARMv8 only.
12225 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
12226 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
12228 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
12229 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
12230 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
12231 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
12232 than TARGET_FPU_ARMV8.
12233 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
12234 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
12235 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
12237 * config/arm/neon.md (neon_vrint): Likewise.
12238 (neon_vcvt): Likewise.
12239 (neon_<fmaxmin_op><mode>): Likewise.
12240 (<fmaxmin><mode>3): Likewise.
12241 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
12242 * config/arm/predicates.md (arm_cond_move_operator): Check against
12243 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
12245 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
12247 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
12248 to top of function.
12250 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12252 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
12253 loop in comment with memset.
12255 2017-07-14 Martin Liska <mliska@suse.cz>
12257 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
12258 * dwarf2out.c (is_java): Remove the function.
12259 (output_pubname): Remove usage of the function.
12260 (lower_bound_default): Remove usage of DW_LANG_Java.
12261 (gen_compile_unit_die): Likewise.
12262 * gcc.c: Remove compiler defaults for .java and .zip files.
12263 * gimple-expr.c (remove_suffix): Change as there's no longer
12264 extension than 4-letter one.
12265 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
12266 (gimplify_save_expr): Likewise.
12267 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
12268 as it's possible even for other languages than Java.
12269 * langhooks.h (struct lang_hooks): Remove Java from a comment.
12270 * lto-opts.c (lto_write_options): Remove reference to Java.
12271 * opts.c (strip_off_ending): Update file extension handling.
12272 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
12273 * tree-eh.c (lower_resx): Likewise.
12274 * tree.c (free_lang_data_in_type): Remove dead code.
12275 (find_decls_types_r): Likewise.
12276 (build_common_builtin_nodes): Remove Java from a comment.
12277 (verify_type): Remove dead code.
12278 * varasm.c (assemble_external): Remove Java from a comment.
12280 2017-07-14 Martin Liska <mliska@suse.cz>
12282 * opts.c (finish_options): Add quotes.
12283 (common_handle_option): Likewise.
12285 2017-07-14 Martin Liska <mliska@suse.cz>
12287 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
12288 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
12289 Remove N_SO_PASCAL.
12290 * dwarf2out.c (lower_bound_default): Do not handle
12292 (gen_compile_unit_die): Likewise.
12293 * gcc.c: Remove default extension binding for GNU Pascal.
12294 * stmt.c: Remove Pascal language from a comment.
12295 * xcoffout.c: Likewise.
12297 2017-07-13 David Malcolm <dmalcolm@redhat.com>
12300 * diagnostic-show-locus.c (fixit_cmp): New function.
12301 (layout::layout): Sort m_fixit_hints.
12302 (column_range::column_range): Assert that the values are valid.
12303 (struct char_span): New struct.
12304 (correction::overwrite): New method.
12305 (struct source_line): New struct.
12306 (line_corrections::add_hint): Add assertions. Reimplement memcpy
12307 calls in terms of classes source_line and char_span, and
12308 correction::overwrite.
12309 (selftest::test_overlapped_fixit_printing_2): New function.
12310 (selftest::diagnostic_show_locus_c_tests): Call it.
12312 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
12314 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
12315 early if there is no lhs.
12317 2017-07-13 Martin Liska <mliska@suse.cz>
12319 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
12320 (gen_reference_type_die): Likewise.
12321 * stor-layout.c: Remove Pascal-related comment.
12323 2017-07-13 Martin Liska <mliska@suse.cz>
12325 * opts.c (finish_options): Add quotes to error messages.
12326 (parse_sanitizer_options): Likewise.
12328 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
12330 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
12332 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
12334 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
12336 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
12338 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
12340 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
12342 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
12345 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
12346 provides the hardware capability bits, define the macro
12347 __BUILTIN_CPU_SUPPORTS__.
12348 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
12349 if GLIBC does not provide the hardware capability bits. Add a
12350 gcc_unreachable call if the built-in cpu function is neither
12351 __builtin_cpu_is nor __builtin_cpu_supports.
12352 (rs6000_get_function_versions_dispatcher): Change the warning
12353 that an old GLIBC is used which does not export the capability
12354 bits to be an error.
12355 * doc/extend.texi (target_clones attribute): Document the
12356 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
12357 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
12358 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
12359 the macros defined by GCC if the newer GLIBC is available.
12361 2017-07-12 Jeff Law <law@redhat.com>
12363 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
12364 remaining includes slightly.
12365 * config/riscv/riscv-builtins.c: Include profile-count.h.
12367 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
12370 * config/avr/avr.c (avr_set_current_function): In diagnostic
12371 messages: Quote keywords and (parts of) identifiers.
12372 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
12375 2017-07-12 Carl Love <cel@us.ibm.com>
12377 * config/rs6000/rs6000-c.c: Add support for built-in functions
12378 vector bool char vec_revb (vector bool char);
12379 vector bool short vec_revb (vector short char);
12380 vector bool int vec_revb (vector bool int);
12381 vector bool long long vec_revb (vector bool long long);
12382 * doc/extend.texi: Update the built-in documentation file for the
12383 new built-in functions.
12385 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12387 * config/s390/s390.md: Remove movcc splitter.
12389 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12391 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
12392 load/store on condition.
12394 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
12397 * config/avr/avr.c (avr_encode_section_info)
12398 [progmem && !TREE_READONLY]: Error if progmem object needs
12401 2017-07-11 Michael Collison <michael.collison@arm.com>
12403 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
12406 2017-07-11 Carl Love <cel@us.ibm.com>
12408 * config/rs6000/rs6000-c.c: Add support for builtins
12409 vector unsigned int vec_parity_lsbb (vector signed int);
12410 vector unsigned int vec_parity_lsbb (vector unsigned int);
12411 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
12412 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
12413 vector unsigned long long vec_parity_lsbb (vector signed long long);
12414 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
12415 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
12416 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
12417 * doc/extend.texi: Update the built-in documentation file for the
12418 new built-in functions.
12420 2017-07-11 David Malcolm <dmalcolm@redhat.com>
12422 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
12423 (layout::m_primary_loc): New field.
12424 (layout::layout): Initialize new field. Move location filtering
12425 logic from here to...
12426 (layout::maybe_add_location_range): ...this new method. Add
12427 support for filtering to just the lines already specified by other
12429 (layout::will_show_line_p): New method.
12430 (gcc_rich_location::add_location_if_nearby): New method.
12431 (selftest::test_add_location_if_nearby): New test function.
12432 (selftest::diagnostic_show_locus_c_tests): Call it.
12433 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
12436 2017-07-11 Tom de Vries <tom@codesourcery.com>
12438 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
12439 (bb_first_real_insn): New function.
12440 (nvptx_single): Add extra initialization of broadcasted condition
12443 2017-07-11 Nathan Sidwell <nathan@acm.org>
12445 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
12447 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
12449 * doc/extend.texi (AVR Function Attributes): Remove weblink to
12450 Binutils doc as TEXI will mess them up.
12451 * doc/invoke.texi (AVR Options): Same here.
12453 2017-07-11 Daniel Cederman <cederman@gaisler.com>
12455 * config/sparc/sparc.opt (mfix-ut700): New option.
12456 (mfix-gr712rc): Likewise.
12457 (sparc_fix_b2bst): New variable.
12458 * doc/invoke.texi (SPARC options): Document them.
12459 (ARM options): Fix warnings.
12460 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
12461 instructions to prevent sequences that can trigger the store-store
12462 errata for certain LEON3FT processors.
12463 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
12464 (sparc_option_override): Set sparc_fix_b2bst appropriately.
12465 * config/sparc/sparc.md (fix_b2bst): New attribute.
12466 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
12468 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
12471 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
12473 (*rsqrtsf2_sse): Ditto.
12475 (div<mode>3): Macroize insn from divdf3 and divsf3
12476 using MODEF mode iterator.
12478 2017-07-10 Martin Sebor <msebor@redhat.com>
12480 PR tree-optimization/80397
12481 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
12482 instead of testing for equality to INTEGER_TYPE.
12484 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
12486 * config.gcc: Remove uclibc from arc target spec.
12488 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
12490 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
12492 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
12495 * lto-wrapper.c (remove_option): New function.
12496 (merge_and_complain): Merge PIC/PIE options more realistically.
12498 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
12500 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
12504 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
12505 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
12506 * config.in: Regenerate.
12507 * configure: Regenerate.
12508 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
12509 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
12510 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
12511 (TARGET_GASISR_PROLOGUES): ...target mask.
12512 * common/config/avr/avr-common.c
12513 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
12514 Set -mgas-isr-prologues.
12515 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
12516 INSERT_PASS_BEFORE for it.
12517 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
12518 * config/avr/avr.c (avr_option_override)
12519 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
12520 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
12521 (avr_attribute_table) <no_gccisr>: Add new function attribute.
12522 (avr_set_current_function) <is_no_gccisr>: Init machine field.
12523 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
12525 (make_avr_pass_pre_proep): New function.
12526 (emit_push_sfr) <treg>: Add argument to function and use it
12527 instead of TMP_REG.
12528 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
12529 and set machine->gasisr.yes.
12530 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
12531 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
12532 __gcc_isr.n_pushed to .L__stack_usage.
12533 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
12534 (avr_asm_final_postscan_insn): ...this new static function.
12535 * config/avr/avr.h (machine_function)
12536 <is_no_gccisr, use_L__stack_usage>: New fields.
12537 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
12538 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
12539 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
12540 (gasisr, *gasisr): New expander and insn.
12541 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
12542 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
12543 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
12545 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
12547 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
12550 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
12552 Move jump-tables out of .text again.
12555 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
12556 (ASM_OUTPUT_ADDR_VEC): New function.
12557 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
12558 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
12559 INSN_ADDRESSes as asm comment.
12560 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
12561 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
12562 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
12563 * config/avr/avr.md (*tablejump): Adjust comment.
12564 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
12565 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
12567 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
12568 (avr_output_addr_vec): New proto.
12569 (avr_log_t) <insn_addresses>: New field.
12571 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
12574 * config/i386/i386.c (ix86_function_arg_advance): Set
12575 outgoing_args_on_stack to true if there are outgoing arguments
12577 (ix86_function_arg): Likewise.
12578 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
12579 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
12580 * config/i386/i386.h (machine_function): Add
12581 outgoing_args_on_stack.
12583 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
12585 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
12586 supporting pthreds.
12587 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
12589 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
12591 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
12592 (REAL_H): Remove $(MACHMODE_H).
12593 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
12595 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
12596 $(MACHMODE_H) and double-int.h.
12597 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
12599 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
12602 2017-07-07 Andrew Pinski <apinski@cavium.com>
12604 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
12605 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
12607 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
12609 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
12610 Add warning if GCC was not configured to link against a GLIBC that
12611 exports the hardware capability bits.
12612 (make_resolver_func): Make resolver function private and not a
12613 COMDAT function. Create the name with clone_function_name instead
12614 of make_unique_name.
12617 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
12618 correct operand in doing the split.
12620 2017-07-07 Carl Love <cel@us.ibm.com>
12622 * config/rs6000/rs6000-c: Add support for built-in function
12623 vector unsigned short vec_pack_to_short_fp32 (vector float,
12625 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
12626 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
12627 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
12628 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
12629 (convert_4f32_8i16): Add define_expand.
12630 * doc/extend.texi: Update the built-in documentation file for the
12631 new built-in function.
12633 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12635 * config/sparc/m8.md: New file.
12636 * config/sparc/sparc.md: Include m8.md.
12638 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12640 * config/sparc/sparc.opt: New option -mvis4b.
12641 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
12642 (sparc_option_override): Handle VIS4B.
12643 (enum sparc_builtins): Define
12644 SPARC_BUILTIN_DICTUNPACK{8,16,32},
12645 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
12646 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
12647 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
12648 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
12649 (check_constant_argument): New function.
12650 (sparc_vis_init_builtins): Define builtins
12651 __builtin_vis_dictunpack{8,16,32},
12652 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
12653 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
12654 __builtin_vis_fpcmpde{8,16,32}shl and
12655 __builtin_vis_fpcmpur{8,16,32}shl.
12656 (sparc_expand_builtin): Check that the constant operands to
12657 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
12658 constant and in range.
12659 * config/sparc/sparc-c.c (sparc_target_macros): Handle
12661 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
12662 (SPARC_IMM5_P): Likewise.
12663 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
12664 (enabled): Handle vis4b.
12665 (UNSPEC_DICTUNPACK): New unspec.
12666 (UNSPEC_FPCMPSHL): Likewise.
12667 (UNSPEC_FPUCMPSHL): Likewise.
12668 (UNSPEC_FPCMPDESHL): Likewise.
12669 (UNSPEC_FPCMPURSHL): Likewise.
12670 (cpu_feature): New CPU feature `vis4b'.
12671 (dictunpack{8,16,32}): New insns.
12672 (FPCSMODE): New mode iterator.
12673 (fpcscond): New code iterator.
12674 (fpcsucond): Likewise.
12675 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
12676 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
12677 (fpcmpde{8,16,32}{si,di}shl): Likewise.
12678 (fpcmpur{8,16,32}{si,di}shl): Likewise.
12679 * config/sparc/constraints.md: Define constraints `q' for unsigned
12680 2-bit integer constants and `t' for unsigned 5-bit integer
12682 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
12684 (imm5_operand_dictunpack16): Likewise.
12685 (imm5_operand_dictunpack32): Likewise.
12686 (imm2_operand): Likewise.
12687 * doc/invoke.texi (SPARC Options): Document -mvis4b.
12688 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
12689 ditunpack* and fpcmp*shl builtins.
12691 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12693 * config.gcc: Handle m8 in --with-{cpu,tune} options.
12694 * config.in: Add HAVE_AS_SPARC6 define.
12695 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
12697 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
12699 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
12700 (CPP_CPU_SPEC): Handle m8.
12701 (ASM_CPU_SPEC): Likewise.
12702 * config/sparc/sparc-opts.h (enum processor_type): Add
12704 * config/sparc/sparc.c (m8_costs): New struct.
12705 (sparc_option_override): Handle TARGET_CPU_m8.
12706 (sparc32_initialize_trampoline): Likewise.
12707 (sparc64_initialize_trampoline): Likewise.
12708 (sparc_issue_rate): Likewise.
12709 (sparc_register_move_cost): Likewise.
12710 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
12711 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
12712 (ASM_CPU64_DEFAULT_SPEC): Likewise.
12713 (CPP_CPU_SPEC): Handle M8.
12714 (ASM_CPU_SPEC): Likewise.
12715 (AS_M8_FLAG): Define.
12716 * config/sparc/sparc.md: Add m8 to the cpu attribute.
12717 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
12718 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
12720 * configure: Regenerate.
12721 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
12724 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12726 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
12728 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
12729 ("*movdi_insn_sp32"): Do not set v3pipe.
12730 ("*movsi_insn"): Likewise.
12731 ("*movdi_insn_sp64"): Likewise.
12732 ("*movsf_insn"): Likewise.
12733 ("*movdf_insn_sp32"): Likewise.
12734 ("*movdf_insn_sp64"): Likewise.
12735 ("*zero_extendsidi2_insn_sp64"): Likewise.
12736 ("*sign_extendsidi2_insn"): Likewise.
12737 ("*mov<VM32:mode>_insn"): Likewise.
12738 ("*mov<VM64:mode>_insn_sp64"): Likewise.
12739 ("*mov<VM64:mode>_insn_sp32"): Likewise.
12740 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
12741 ("<vlop:code><VL:mode>3"): Likewise.
12742 ("*not_<vlop:code><VL:mode>3"): Likewise.
12743 ("*nand<VL:mode>_vis"): Likewise.
12744 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
12745 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
12746 ("one_cmpl<VL:mode>2"): Likewise.
12747 ("faligndata<VM64:mode>_vis"): Likewise.
12748 ("alignaddrsi_vis"): Likewise.
12749 ("alignaddrdi_vis"): Likweise.
12750 ("alignaddrlsi_vis"): Likewise.
12751 ("alignaddrldi_vis"): Likewise.
12752 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
12753 ("bmaskdi_vis"): Likewise.
12754 ("bmasksi_vis"): Likewise.
12755 ("bshuffle<VM64:mode>_vis"): Likewise.
12756 ("cmask8<P:mode>_vis"): Likewise.
12757 ("cmask16<P:mode>_vis"): Likewise.
12758 ("cmask32<P:mode>_vis"): Likewise.
12759 ("pdistn<P:mode>_vis"): Likewise.
12760 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
12762 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12764 * config/sparc/sparc.md ("subtype"): New insn attribute.
12765 ("*wrgsr_sp64"): Set insn subtype.
12766 ("*rdgsr_sp64"): Likewise.
12767 ("alignaddrsi_vis"): Likewise.
12768 ("alignaddrdi_vis"): Likewise.
12769 ("alignaddrlsi_vis"): Likewise.
12770 ("alignaddrldi_vis"): Likewise.
12771 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
12772 ("fexpand_vis"): Likewise.
12773 ("fpmerge_vis"): Likewise.
12774 ("faligndata<VM64:mode>_vis"): Likewise.
12775 ("bshuffle<VM64:mode>_vis"): Likewise.
12776 ("cmask8<P:mode>_vis"): Likewise.
12777 ("cmask16<P:mode>_vis"): Likewise.
12778 ("cmask32<P:mode>_vis"): Likewise.
12779 ("fchksm16_vis"): Likewise.
12780 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
12781 ("fmean16_vis"): Likewise.
12782 ("fp<plusminus_insn>64_vis"): Likewise.
12783 ("<plusminus_insn>v8qi3"): Likewise.
12784 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
12785 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
12786 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
12787 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
12788 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
12789 ("*movqi_insn"): Likewise.
12790 ("*movhi_insn"): Likewise.
12791 ("*movsi_insn"): Likewise.
12792 ("movsi_pic_gotdata_op"): Likewise.
12793 ("*movdi_insn_sp32"): Likewise.
12794 ("*movdi_insn_sp64"): Likewise.
12795 ("movdi_pic_gotdata_op"): Likewise.
12796 ("*movsf_insn"): Likewise.
12797 ("*movdf_insn_sp32"): Likewise.
12798 ("*movdf_insn_sp64"): Likewise.
12799 ("*zero_extendhisi2_insn"): Likewise.
12800 ("*zero_extendqihi2_insn"): Likewise.
12801 ("*zero_extendqisi2_insn"): Likewise.
12802 ("*zero_extendqidi2_insn"): Likewise.
12803 ("*zero_extendhidi2_insn"): Likewise.
12804 ("*zero_extendsidi2_insn_sp64"): Likewise.
12805 ("ldfsr"): Likewise.
12806 ("prefetch_64"): Likewise.
12807 ("prefetch_32"): Likewise.
12808 ("tie_ld32"): Likewise.
12809 ("tie_ld64"): Likewise.
12810 ("*tldo_ldub_sp32"): Likewise.
12811 ("*tldo_ldub1_sp32"): Likewise.
12812 ("*tldo_ldub2_sp32"): Likewise.
12813 ("*tldo_ldub_sp64"): Likewise.
12814 ("*tldo_ldub1_sp64"): Likewise.
12815 ("*tldo_ldub2_sp64"): Likewise.
12816 ("*tldo_ldub3_sp64"): Likewise.
12817 ("*tldo_lduh_sp32"): Likewise.
12818 ("*tldo_lduh1_sp32"): Likewise.
12819 ("*tldo_lduh_sp64"): Likewise.
12820 ("*tldo_lduh1_sp64"): Likewise.
12821 ("*tldo_lduh2_sp64"): Likewise.
12822 ("*tldo_lduw_sp32"): Likewise.
12823 ("*tldo_lduw_sp64"): Likewise.
12824 ("*tldo_lduw1_sp64"): Likewise.
12825 ("*tldo_ldx_sp64"): Likewise.
12826 ("*mov<VM32:mode>_insn"): Likewise.
12827 ("*mov<VM64:mode>_insn_sp64"): Likewise.
12828 ("*mov<VM64:mode>_insn_sp32"): Likewise.
12830 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12832 * config/sparc/sparc.md ("type"): New insn type viscmp.
12833 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
12835 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
12836 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
12837 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
12838 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
12840 ("n7_vis_logical_11cycle"): Likewise.
12841 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
12842 * config/sparc/niagara2.md ("niag3_vis": Likewise.
12843 * config/sparc/niagara.md ("niag_vis"): Likewise.
12844 * config/sparc/ultra3.md ("us3_fga"): Likewise.
12845 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
12847 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12849 * config/sparc/sparc.md: New instruction type `bmask'.
12850 (bmaskdi_vis): Use the `bmask' type.
12851 (bmasksi_vis): Likewise.
12852 * config/sparc/ultra3.md (us3_array): Likewise.
12853 * config/sparc/niagara7.md (n7_array): Likewise.
12854 * config/sparc/niagara4.md (n4_array): Likewise.
12855 * config/sparc/niagara2.md (niag2_vis): Likewise.
12856 (niag3_vis): Likewise.
12857 * config/sparc/niagara.md (niag_vis): Likewise.
12859 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12861 * ipa-comdats.c: Remove optimize check from gate.
12862 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
12863 for functions not optimized.
12864 (ipa_fn_summary_read): Skip optimize check.
12865 (ipa_fn_summary_write): Likewise.
12866 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
12868 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
12870 (can_inline_edge_p): Check flag_pcc_struct_return for match.
12871 (check_callers): Give up on caller which is not optimized.
12872 (inline_small_functions): Likewise.
12873 (ipa_inline): Do not give up when not optimizing.
12874 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
12875 away unoptimizes cdtors.
12876 (whole_program_function_and_variable_visibility): Do
12877 ipa_discover_readonly_nonaddressable_vars in LTO mode.
12878 * ipa.c (process_references): Do not check optimize.
12879 (symbol_table::remove_unreachable_nodes): Update optimize check.
12880 (set_writeonly_bit): Update optimize check.
12881 (pass_ipa_cdtor_merge::gate): Do not check optimize.
12882 (pass_ipa_single_use::gate): Remove.
12884 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12886 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
12887 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
12888 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
12889 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
12890 permute_load, permute_store, adjust_extract, adjust_splat,
12891 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
12892 replace_swap_with_copy, dump_swap_insn_table,
12893 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
12894 recombine_lvx_pattern, recombine_stvx_pattern,
12895 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
12896 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
12897 to file rs6000-p8swap.c.
12898 * config/rs6000/rs6000-p8swap.c: New file.
12899 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
12900 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
12901 and rs6000*-*-* targets.
12903 2017-07-06 David Malcolm <dmalcolm@redhat.com>
12905 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
12907 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12909 * lto-wrapper.c (merge_and_complain): Do not merge
12910 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
12911 fsigned_zeros, ftrapping_math, fwrapv.
12912 (append_compiler_options): Do not track these options.
12913 (append_linker_options): Likewie
12915 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12917 * cgraphunit.c (cgraph_node::finalize_function): When
12918 !flag_toplevel_reorde set no_reorder flag.
12919 (varpool_node::finalize_decl): Likewise.
12920 (symbol_table::compile): Drop no toplevel reorder path.
12922 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12924 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
12925 edges; zero probability is not better than uninitialized.
12927 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
12929 * asan.h (asan_sanitize_allocas_p): Declare.
12930 * asan.c (asan_sanitize_allocas_p): New function.
12931 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
12932 (handle_builtin_alloca): Likewise.
12933 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
12934 if !asan_sanitize_allocas_p.
12935 * params.def (asan-instrument-allocas): Add new option.
12936 * params.h (ASAN_PROTECT_ALLOCAS): Define.
12937 * opts.c (common_handle_option): Disable allocas sanitization for
12940 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
12942 * asan.c: Include gimple-fold.h.
12943 (get_last_alloca_addr): New function.
12944 (handle_builtin_stackrestore): Likewise.
12945 (handle_builtin_alloca): Likewise.
12946 (asan_emit_allocas_unpoison): Likewise.
12947 (get_mem_refs_of_builtin_call): Add new parameter, remove const
12948 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
12949 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
12950 (instrument_builtin_call): Pass gimple iterator to
12951 get_mem_refs_of_builtin_call.
12952 (last_alloca_addr): New global.
12953 * asan.h (asan_emit_allocas_unpoison): Declare.
12954 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
12955 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
12956 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
12957 if function calls alloca.
12958 * gimple-fold.c (replace_call_with_value): Remove static keyword.
12959 * gimple-fold.h (replace_call_with_value): Declare.
12960 * internal-fn.c: Include asan.h.
12961 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
12962 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
12964 2017-07-06 David Malcolm <dmalcolm@redhat.com>
12966 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
12967 (C_SELFTEST_FLAGS): New.
12968 (CPP_SELFTEST_FLAGS): New.
12969 (SELFTEST_DEPS): New, from deps of s-selftest.
12970 (C_SELFTEST_DEPS): New, from deps of s-selftest.
12971 (CPP_SELFTEST_DEPS): New.
12972 (selftest): Add dependency on s-selftest-c++.
12973 (s-selftest): Rename to...
12974 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
12975 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
12976 than SELFTEST_FLAGS.
12977 (selftest-gdb): Rename to...
12978 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
12980 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
12981 (selftest-valgrind): Rename to...
12982 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
12984 (selftest-valgrind): Reintroduce as an alias for
12985 selftest-c-valgrind.
12986 (s-selftest-c++): New.
12987 (selftest-c++-gdb): New.
12988 (selftest-c++-valgrind): New.
12990 2017-07-06 Olivier Hainque <hainque@adacore.com>
12992 * gcc.c (process_command): When deciding if undefined variables
12993 should be ignored when processing specs, accept "gcc -v" as well.
12995 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12997 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
12998 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
13000 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
13002 * config/arm/arm-cpus.in (armv8-r): Add new entry.
13003 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
13004 * config/arm/arm-tables.opt: Regenerate.
13005 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
13007 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
13009 2017-07-06 Carl Love <cel@us.ibm.com>
13011 * ChangeLog: Clean up from mid air collision
13013 2017-07-06 Carl Love <cel@us.ibm.com>
13015 * config/rs6000/rs6000-c.c: Add support for built-in functions
13016 vector signed int vec_subc (vector signed int, vector signed int);
13017 vector signed __int128 vec_subc (vector signed __int128,
13018 vector signed __int128);
13019 vector unsigned __int128 vec_subc (vector unsigned __int128,
13020 vector unsigned __int128);
13021 vector signed int vec_sube (vector signed int, vector signed int,
13022 vector signed int);
13023 vector unsigned int vec_sube (vector unsigned int,
13024 vector unsigned int,
13025 vector unsigned int);
13026 vector signed __int128 vec_sube (vector signed __int128,
13027 vector signed __int128,
13028 vector signed__int128);
13029 vector unsigned __int128 vec_sube (vector unsigned __int128,
13030 vector unsigned __int128,
13031 vector unsigned __int128);
13032 vector signed int vec_subec (vector signed int, vector signed int,
13033 vector signed int);
13034 vector unsigned int vec_subec (vector unsigned int,
13035 vector unsigned int,
13036 vector unsigned int);
13037 vector signed __int128 vec_subec (vector signed __int128,
13038 vector signed __int128,
13039 vector signed__int128);
13040 vector unsigned __int128 vec_subec (vector unsigned __int128,
13041 vector unsigned __int128,
13042 vector unsigned __int128);
13043 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
13044 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
13045 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
13046 BU_ALTIVEC_OVERLOAD_X definitions.
13047 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
13048 * doc/extend.texi: Update the built-in documentation file for the new
13049 built-in functions.
13051 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13054 * diagnostic-show-locus.c (layout::layout): Use start and finish
13055 spelling location for the start and finish of each range.
13056 * genmatch.c (linemap_client_expand_location_to_spelling_point):
13057 Add unused aspect param.
13058 * input.c (expand_location_1): Add "aspect" param, and use it
13059 to access the correct part of the location.
13060 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
13062 (expand_location_to_spelling_point): Likewise.
13063 (linemap_client_expand_location_to_spelling_point): Add "aspect"
13064 param, and pass it to expand_location_1.
13066 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
13068 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
13069 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
13070 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
13071 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
13072 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
13073 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
13074 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
13075 _mm_maskz_getmant_ss): New intrinsics.
13076 (__builtin_ia32_getexpss128_mask): Changed to ...
13077 __builtin_ia32_getexpss128_round ... this.
13078 (__builtin_ia32_getexpsd128_mask): Changed to ...
13079 __builtin_ia32_getexpsd128_round ... this.
13080 * config/i386/i386-builtin-types.def
13081 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
13082 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
13083 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
13084 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
13085 __builtin_ia32_getmantss_mask_round): New builtins.
13086 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
13087 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
13088 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
13089 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
13090 * config/i386/sse.md
13091 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
13092 avx512f_sgetexp<mode><mask_scalar_name>
13093 <round_saeonly_scalar_name> ... this.
13094 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
13095 %0, %1, %2<round_saeonly_op3>}): Changed to ...
13096 vgetexp<ssescalarmodesuffix>
13097 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
13098 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
13099 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
13100 avx512f_vgetmant<mode><mask_scalar_name>
13101 <round_saeonly_scalar_name> ... this.
13102 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
13103 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
13104 vgetmant<ssescalarmodesuffix>
13105 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
13106 %0<mask_scalar_operand4>, %1, %2
13107 <round_saeonly_scalar_mask_op4>, %3} ... this.
13108 * config/i386/subst.md (mask_scalar_operand4,
13109 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
13110 round_saeonly_scalar_nimm_predicate): New subst attributes.
13112 2017-07-06 Julia Koval <julia.koval@intel.com>
13114 * config/i386/i386.c (ix86_erase_embedded_rounding):
13115 Remove code for old rounding pattern.
13117 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
13119 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
13121 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
13123 * doc/sourcebuild.texi (Test Directives, Variants of
13124 dg-require-support): Add documentation for dg-require-stack-check.
13126 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
13128 * config/i386/subst.md (mask_scalar, round_scalar,
13129 round_saeonly_scalar): New meta-templates.
13130 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
13131 round_scalar_mask_operand3, round_scalar_mask_op3,
13132 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
13133 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
13134 round_saeonly_scalar_constraint,
13135 round_saeonly_scalar_prefix): New subst attribute.
13136 * config/i386/sse.md
13137 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
13138 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
13139 <round_scalar_name> ... this.
13140 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
13141 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
13142 <round_scalar_name> ... this.
13143 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
13144 <sse>_vm<code><mode>3<mask_scalar_name>
13145 <round_saeonly_scalar_name> ... this.
13146 (v<plusminus_mnemonic><ssescalarmodesuffix>
13147 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
13148 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
13149 v<plusminus_mnemonic><ssescalarmodesuffix>
13150 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
13151 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
13152 (v<multdiv_mnemonic><ssescalarmodesuffix>
13153 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
13154 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
13155 v<multdiv_mnemonic><ssescalarmodesuffix>
13156 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
13157 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
13158 (v<maxmin_float><ssescalarmodesuffix>
13159 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
13160 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
13161 v<maxmin_float><ssescalarmodesuffix>
13162 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
13163 %0<mask_scalar_operand3>, %1, %<iptr>2
13164 <round_saeonly_scalar_mask_op3>} ... this.
13166 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
13168 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
13169 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
13171 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
13172 Alan Hayward <alan.hayward@arm.com>
13173 David Sherwood <david.sherwood@arm.com>
13175 * combine.c (simplify_if_then_else): Remove "enum" before
13177 * compare-elim.c (can_eliminate_compare): Likewise.
13178 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
13180 (aarch64_lookup_simd_builtin_type): Likewise.
13181 (aarch64_simd_builtin_type): Likewise.
13182 (aarch64_init_simd_builtin_types): Likewise.
13183 (aarch64_simd_expand_args): Likewise.
13184 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
13186 (aarch64_reverse_mask): Likewise.
13187 (aarch64_simd_emit_reg_reg_move): Likewise.
13188 (aarch64_gen_adjusted_ldpstp): Likewise.
13189 (aarch64_ccmp_mode_to_code): Likewise.
13190 (aarch64_operands_ok_for_ldpstp): Likewise.
13191 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13192 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
13194 (aarch64_min_divisions_for_recip_mul): Likewise.
13195 (aarch64_reassociation_width): Likewise.
13196 (aarch64_get_condition_code_1): Likewise.
13197 (aarch64_simd_emit_reg_reg_move): Likewise.
13198 (aarch64_simd_attr_length_rglist): Likewise.
13199 (aarch64_reverse_mask): Likewise.
13200 (aarch64_operands_ok_for_ldpstp): Likewise.
13201 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13202 (aarch64_gen_adjusted_ldpstp): Likewise.
13203 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
13205 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
13206 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
13207 (arm_lookup_simd_builtin_type): Likewise.
13208 (arm_simd_builtin_type): Likewise.
13209 (arm_init_simd_builtin_types): Likewise.
13210 (arm_expand_builtin_args): Likewise.
13211 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
13212 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
13213 (ft32_setup_incoming_varargs): Likewise.
13214 (ft32_function_arg): Likewise.
13215 (ft32_function_arg_advance): Likewise.
13216 (ft32_pass_by_reference): Likewise.
13217 (ft32_arg_partial_bytes): Likewise.
13218 (ft32_valid_pointer_mode): Likewise.
13219 (ft32_addr_space_pointer_mode): Likewise.
13220 (ft32_addr_space_legitimate_address_p): Likewise.
13221 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
13223 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
13224 (ix86_emit_outlined_ms2sysv_restore): Likewise.
13225 (iamcu_alignment): Likewise.
13226 (canonicalize_vector_int_perm): Likewise.
13227 (ix86_noce_conversion_profitable_p): Likewise.
13228 (ix86_mpx_bound_mode): Likewise.
13229 (ix86_operands_ok_for_move_multiple): Likewise.
13230 * config/microblaze/microblaze-protos.h
13231 (microblaze_expand_conditional_branch_reg): Likewise.
13232 * config/microblaze/microblaze.c
13233 (microblaze_expand_conditional_branch_reg): Likewise.
13234 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
13236 (rs6000_reassociation_width): Likewise.
13237 (rs6000_invalid_binary_op): Likewise.
13238 (fusion_p9_p): Likewise.
13239 (emit_fusion_p9_load): Likewise.
13240 (emit_fusion_p9_store): Likewise.
13241 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
13243 (riscv_hard_regno_mode_ok_p): Likewise.
13244 (riscv_address_insns): Likewise.
13245 (riscv_split_symbol): Likewise.
13246 (riscv_legitimize_move): Likewise.
13247 (riscv_function_value): Likewise.
13248 (riscv_hard_regno_nregs): Likewise.
13249 (riscv_expand_builtin): Likewise.
13250 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
13251 (riscv_build_integer): Likewise.
13252 (riscv_split_integer): Likewise.
13253 (riscv_legitimate_constant_p): Likewise.
13254 (riscv_cannot_force_const_mem): Likewise.
13255 (riscv_regno_mode_ok_for_base_p): Likewise.
13256 (riscv_valid_base_register_p): Likewise.
13257 (riscv_valid_offset_p): Likewise.
13258 (riscv_valid_lo_sum_p): Likewise.
13259 (riscv_classify_address): Likewise.
13260 (riscv_legitimate_address_p): Likewise.
13261 (riscv_address_insns): Likewise.
13262 (riscv_load_store_insns): Likewise.
13263 (riscv_force_binary): Likewise.
13264 (riscv_split_symbol): Likewise.
13265 (riscv_force_address): Likewise.
13266 (riscv_legitimize_address): Likewise.
13267 (riscv_move_integer): Likewise.
13268 (riscv_legitimize_const_move): Likewise.
13269 (riscv_legitimize_move): Likewise.
13270 (riscv_address_cost): Likewise.
13271 (riscv_subword): Likewise.
13272 (riscv_output_move): Likewise.
13273 (riscv_canonicalize_int_order_test): Likewise.
13274 (riscv_emit_int_order_test): Likewise.
13275 (riscv_function_arg_boundary): Likewise.
13276 (riscv_pass_mode_in_fpr_p): Likewise.
13277 (riscv_pass_fpr_single): Likewise.
13278 (riscv_pass_fpr_pair): Likewise.
13279 (riscv_get_arg_info): Likewise.
13280 (riscv_function_arg): Likewise.
13281 (riscv_function_arg_advance): Likewise.
13282 (riscv_arg_partial_bytes): Likewise.
13283 (riscv_function_value): Likewise.
13284 (riscv_pass_by_reference): Likewise.
13285 (riscv_setup_incoming_varargs): Likewise.
13286 (riscv_print_operand): Likewise.
13287 (riscv_elf_select_rtx_section): Likewise.
13288 (riscv_save_restore_reg): Likewise.
13289 (riscv_for_each_saved_reg): Likewise.
13290 (riscv_register_move_cost): Likewise.
13291 (riscv_hard_regno_mode_ok_p): Likewise.
13292 (riscv_hard_regno_nregs): Likewise.
13293 (riscv_class_max_nregs): Likewise.
13294 (riscv_memory_move_cost): Likewise.
13295 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
13296 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
13297 (rl78_addr_space_address_mode): Likewise.
13298 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13300 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13301 (rs6000_reassociation_width): Likewise.
13302 (rs6000_invalid_binary_op): Likewise.
13303 (fusion_p9_p): Likewise.
13304 (emit_fusion_p9_load): Likewise.
13305 (emit_fusion_p9_store): Likewise.
13306 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
13307 (ok_for_simple_move_operands): Likewise.
13308 (ok_for_simple_move_strict_operands): Likewise.
13309 (ok_for_simple_arith_logic_operands): Likewise.
13310 (visium_legitimize_reload_address): Likewise.
13311 (visium_select_cc_mode): Likewise.
13312 (output_cbranch): Likewise.
13313 (visium_split_double_move): Likewise.
13314 (visium_expand_copysign): Likewise.
13315 (visium_expand_int_cstore): Likewise.
13316 (visium_expand_fp_cstore): Likewise.
13317 * config/visium/visium.c (visium_pass_by_reference): Likewise.
13318 (visium_function_arg): Likewise.
13319 (visium_function_arg_advance): Likewise.
13320 (visium_libcall_value): Likewise.
13321 (visium_setup_incoming_varargs): Likewise.
13322 (visium_legitimate_constant_p): Likewise.
13323 (visium_legitimate_address_p): Likewise.
13324 (visium_legitimize_address): Likewise.
13325 (visium_secondary_reload): Likewise.
13326 (visium_register_move_cost): Likewise.
13327 (visium_memory_move_cost): Likewise.
13328 (prepare_move_operands): Likewise.
13329 (ok_for_simple_move_operands): Likewise.
13330 (ok_for_simple_move_strict_operands): Likewise.
13331 (ok_for_simple_arith_logic_operands): Likewise.
13332 (visium_function_value_1): Likewise.
13333 (rtx_ok_for_offset_p): Likewise.
13334 (visium_legitimize_reload_address): Likewise.
13335 (visium_split_double_move): Likewise.
13336 (visium_expand_copysign): Likewise.
13337 (visium_expand_int_cstore): Likewise.
13338 (visium_expand_fp_cstore): Likewise.
13339 (visium_split_cstore): Likewise.
13340 (visium_select_cc_mode): Likewise.
13341 (visium_split_cbranch): Likewise.
13342 (output_cbranch): Likewise.
13343 (visium_print_operand_address): Likewise.
13344 * expmed.c (flip_storage_order): Likewise.
13345 * expmed.h (emit_cstore): Likewise.
13346 (flip_storage_order): Likewise.
13347 * genrecog.c (validate_pattern): Likewise.
13348 * hsa-gen.c (gen_hsa_addr): Likewise.
13349 * internal-fn.c (expand_arith_overflow): Likewise.
13350 * ira-color.c (allocno_copy_cost_saving): Likewise.
13351 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13352 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
13353 (process_invariant_for_inheritance): Likewise.
13354 * lra-eliminations.c (move_plus_up): Likewise.
13355 * omp-low.c (lower_oacc_reductions): Likewise.
13356 * simplify-rtx.c (simplify_subreg): Likewise.
13357 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
13358 (TARGET_CHKP_BOUND_MODE): Likewise..
13359 * targhooks.c (default_chkp_bound_mode): Likewise.
13360 (default_setup_incoming_vararg_bounds): Likewise.
13361 * targhooks.h (default_chkp_bound_mode): Likewise.
13362 (default_setup_incoming_vararg_bounds): Likewise.
13363 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
13364 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
13365 (have_whole_vector_shift): Likewise.
13366 * tree-vect-stmts.c (vectorizable_load): Likewise.
13367 * doc/tm.texi: Regenerate.
13369 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
13371 Graceful degrade if Binutils PR21472 is not available.
13374 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
13375 .rodata in flash test fails.
13376 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
13377 * confgure: Regenerate.
13378 * config.in: Regenerate.
13379 * config/avr/avr.c (avr_asm_named_section)
13380 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
13381 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
13382 (avr_asm_init_sections): Same.
13384 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13386 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
13387 (fma<VH:mode>4_intrinsic): Likewise.
13388 (*fmsub<VCVTF:mode>4): Likewise.
13389 (*fmsub<VH:mode>4_intrinsic): Likewise.
13391 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
13394 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
13395 Don't depend on "optimize > 0".
13396 (out_movhi_r_mr, out_movqi_mr_r): Same.
13397 (out_movhi_mr_r, out_movqi_r_mr): Same.
13398 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
13399 io_address_operand on "optimize > 0".
13401 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13403 * tree-loop-distribution.c: Add general explanantion on the pass.
13404 (generate_loops_for_partition): Mark distributed loop.
13405 (pg_add_dependence_edges): New parameter. Handle alias data
13406 dependence specially and record it in the parameter if asked.
13407 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
13408 (init_partition_graph_vertices, add_partition_graph_edge): New.
13409 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
13410 (free_partition_graph_vdata, build_partition_graph): New.
13411 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
13412 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
13413 (data_ref_segment_size, latch_dominated_by_data_ref): New.
13414 (compute_alias_check_pairs, version_loop_by_alias_check): New.
13415 (version_for_distribution_p, finalize_partitions): New.
13416 (distribute_loop): Handle alias data dependence specially. Factor
13417 out loop fusion code as functions and call these functions.
13419 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13421 * tree-loop-distribution.c (classify_partition): New parameter and
13422 better handle reduction statement.
13423 (rdg_build_partitions): Revise comment.
13424 (distribute_loop): Compute statements in all partitions and pass it
13425 to classify_partition.
13427 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13429 * tree-loop-distribution.c (enum partition_type): New.
13430 (struct partition): New field type.
13431 (partition_merge_into): Add parameter. Update partition type.
13432 (data_dep_in_cycle_p, update_type_for_merge): New functions.
13433 (build_rdg_partition_for_vertex): Compute partition type.
13434 (rdg_build_partitions): Dump partition type.
13435 (distribute_loop): Update calls to partition_merge_into.
13437 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13439 * tree-loop-distribution.c (struct ddr_hasher): New.
13440 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
13442 (classify_partition): Call get_data_dependence.
13443 (pg_add_dependence_edges): Ditto.
13444 (distribute_loop): Release data dependence hash table.
13446 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13448 * tree-loop-distribution.c (ref_base_address): Delete.
13449 (similar_memory_accesses): Rename ...
13450 (share_memory_accesses): ... to this. Check if partitions access
13451 the same memory reference.
13452 (distribute_loop): Call share_memory_accesses.
13454 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13456 * tree-loop-distribution.c (struct partition): New field recording
13457 its data reference.
13458 (partition_alloc, partition_free): Init and release data refs.
13459 (partition_merge_into): Merge data refs.
13460 (build_rdg_partition_for_vertex): Collect data refs for partition.
13461 (pg_add_dependence_edges): Change parameters from vector to bitmap.
13463 (distribute_loop): Remve data refs from vertice data of partition
13466 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13468 * tree-loop-distribution.c (params.h): Include header file.
13469 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
13470 (datarefs_vec): New global var.
13471 (create_rdg_vertices): Use datarefs_vec directly.
13472 (free_rdg): Don't free data references.
13473 (build_rdg): Update use. Don't free data references.
13474 (distribute_loop): Compute global variable for data references.
13475 Bail out if there are too many data references.
13477 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13479 * tree-loop-distribution.c (loop_nest): New global var.
13480 (build_rdg): Use loop directly, rather than loop nest.
13481 (pg_add_dependence_edges): Remove loop nest parameter. Use global
13483 (distribute_loop): Compute global variable loop nest. Update use.
13485 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13487 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
13488 (partition_merge_into): New parameter. Dump reason for fusion.
13489 (distribute_loop): Update use of partition_merge_into.
13491 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13493 * tree-loop-distribution.c (bb_top_order_index): New.
13494 (bb_top_order_index_size, bb_top_order_cmp): New.
13495 (stmts_from_loop): Use topological order.
13496 (pass_loop_distribution::execute): Compute and release topological
13497 order for basic blocks.
13499 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13501 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
13504 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13506 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
13507 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
13508 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
13509 * internal-fn.def (LOOP_DIST_ALIAS): New.
13510 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
13511 (fold_loop_internal_call): ... this.
13512 (vect_loop_dist_alias_call): New function.
13513 (set_uid_loop_bbs): Call fold_loop_internal_call.
13514 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
13517 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
13520 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
13521 Require dead FLAGS_REG at the beginning of a peephole.
13523 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
13526 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
13527 arguments in the call to __builtin_ia32_sbb_u32.
13528 (_subborrow_u64): Swap _X and _Y arguments in the call to
13529 __builtin_ia32_sbb_u64.
13531 2017-07-04 Jakub Jelinek <jakub@redhat.com>
13534 * tree-vrp.c (compare_assert_loc): Turn into a function template
13535 with stable template parameter. Only test if a->e is NULL,
13536 !a->e == !b->e has been verified already. Use e == NULL or
13537 e != NULL instead of e or ! e tests. If stable is true, don't use
13538 iterative_hash_expr, on the other side allow a or b or both NULL
13539 and sort the NULLs last.
13540 (process_assert_insertions): Sort using compare_assert_loc<false>
13541 instead of compare_assert_loc, later sort using
13542 compare_assert_loc<true> before calling process_assert_insertions_for
13543 in a loop. Use break instead of continue once seen NULL pointer.
13545 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
13547 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
13548 Cortex-R7 and Cortex-R8 processors.
13550 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
13552 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
13553 uninitialized while src is not.
13555 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
13557 * common/config/arm/arm-common.c: Adjust include path for
13559 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
13560 (arm-cpu.h): Create in build directory. Adjust dependency rules.
13561 (arm-cpu-data.h): Likewise.
13562 (arm-cpu-cdata.h): Likewise.
13563 * config/arm/arm-cpu.h: Delete.
13564 * config/arm/arm-cpu-cdata.h: Delete.
13565 * config/arm/arm-cpu-data.h: Delete.
13567 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
13569 * config/arm/arm-cpus.in (cortex-a55): New.
13570 (cortex-a75): Likewise.
13571 (cortex-a75.cortex-a55): Likewise.
13572 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
13574 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
13575 * config/arm/arm-cpu-cdata.h: Regenerate.
13576 * config/arm/arm-cpu-data.h: Regenerate.
13577 * config/arm/arm-cpu.h: Regenerate.
13578 * config/arm/arm-tables.opt: Regenerate.
13579 * config/arm/arm-tune.md: Regenerate.
13581 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
13583 * haifa-sched.c (sched_create_recovery_edges): Update profile.
13585 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
13587 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
13590 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
13592 PR tree-optimization/81292
13593 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
13594 full_string_p, also call adjust_related_strinfos if the adjustment
13595 is simple, otherwise invalidate related strinfos.
13597 2017-07-04 Martin Liska <mliska@suse.cz>
13600 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
13601 newly created variable as DECL_IGNORED_P.
13603 2017-07-04 Martin Liska <mliska@suse.cz>
13606 * ipa-inline.c (inline_small_functions):
13607 Use xstrdup_for_dump.
13609 2017-07-04 Tom de Vries <tom@codesourcery.com>
13611 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
13613 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
13616 * config/darwin.c (darwin_function_switched_text_sections):
13619 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
13621 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
13623 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
13625 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
13627 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13629 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
13630 min_profitable_iters, and th as inclusive lower bounds.
13631 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
13632 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
13633 for min_profitable_iters and min_profitable_estimate.
13634 (vect_transform_loop): Treat th as an inclusive lower bound.
13635 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
13637 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
13640 * config/darwin.c (darwin_function_switched_text_sections):
13641 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
13642 in two pieces, and suppress the use of buf.
13644 2017-07-03 Nathan Sidwell <nathan@acm.org>
13646 * hash-table.h (hash_table_mod1): Fix indentation.
13648 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13650 PR middle-end/81290
13651 * predict.c (force_edge_cold): Be more careful about propagation
13653 * profile-count.h (profile_probability::guessed,
13654 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
13656 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
13658 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
13660 * doc/invoke.texi (rcpc architecture extension): Document it.
13662 2017-07-03 Richard Biener <rguenther@suse.de>
13664 PR tree-optimization/60510
13665 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
13666 the scalar reduction PHI and use it.
13667 (vectorizable_reduction): Properly guard the single_defuse_cycle
13668 path for non-SLP reduction chains where we cannot use it.
13669 Rework reduc_def/index and vector type deduction. Rework
13670 vector operand gathering during reduction op code-gen.
13671 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
13672 chains dissolve the chain and leave it to non-SLP reduction
13675 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13677 * tree-data-ref.h (dr_alignment): Declare.
13678 * tree-data-ref.c (dr_alignment): New function.
13679 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
13680 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
13682 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
13684 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13686 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
13687 and base_misalignment fields.
13688 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
13689 * tree-data-ref.c: Include builtins.h.
13690 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
13691 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
13692 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
13693 * tree-vect-data-refs.c: Include tree-cfg.h.
13694 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
13695 fields instead of calculating an alignment here.
13696 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
13697 innermost_loop_behavior fields.
13699 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13701 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
13703 (DR_STEP_ALIGNMENT): New macro.
13704 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
13705 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
13706 (create_data_ref): Print it.
13707 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
13708 to tell whether the step preserves vector (mis)alignment.
13709 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
13710 Move the check for an integer step and generalise to all INTEGER_CST.
13711 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
13712 Print the outer step alignment.
13714 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13716 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
13717 with offset_alignment.
13718 (DR_ALIGNED_TO): Delete.
13719 (DR_OFFSET_ALIGNMENT): New macro.
13720 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
13721 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
13722 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
13723 (create_data_ref): Likewise.
13724 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
13725 (vect_analyze_data_refs): Likewise.
13726 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
13727 creating dummy innermost behavior.
13729 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13731 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
13732 with a "innermost_loop_behavior *" and refeence tree.
13733 * tree-data-ref.c (dr_analyze_innermost): Likewise.
13734 (create_data_ref): Update call accordingly.
13735 * tree-predcom.c (find_looparound_phi): Likewise.
13737 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13739 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
13740 fields with dr_wrt_vec_loop.
13741 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
13742 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
13743 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
13744 (vect_dr_behavior): New function.
13745 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
13746 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
13747 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
13748 track whether the step preserves the misalignment.
13749 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
13750 Use vect_dr_behavior.
13751 (vect_setup_realignment): Update call accordingly.
13752 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
13753 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
13754 call to vect_create_addr_base_for_vector_ref.
13755 (vect_create_cond_for_align_checks): Likewise.
13756 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
13757 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
13758 (vect_recog_mask_conversion_pattern): Likewise.
13759 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
13760 (new_stmt_vec_info): Remove redundant zeroing.
13762 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
13764 * common/config/arm/arm-common.c (arm_be8_option): New function.
13765 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
13766 (ISA_ARMv6): Add isa_bit_be8.
13767 * config/arm/arm.h (arm_be8_option): Add prototype.
13768 (BE8_SPEC_FUNCTION): New define.
13769 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
13770 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
13771 (mlittle-endian): Similarly.
13772 (mbe8, mbe32): New options.
13773 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
13774 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
13776 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13778 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
13780 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13782 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
13783 (cleanup_tree_cfg_bb): Use it.
13784 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
13786 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
13788 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13791 * loop-doloop.c (add_test): Update profile.
13793 2017-07-03 Martin Liska <mliska@suse.cz>
13796 * sanopt.c (rewrite_usage_of_param): New function.
13797 (sanitize_rewrite_addressable_params): Likewise.
13798 (pass_sanopt::execute): Call rewrite_usage_of_param.
13800 2017-07-03 Richard Biener <rguenther@suse.de>
13802 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
13803 back to using VIEW_CONVERT_EXPR.
13805 2017-07-03 Martin Liska <mliska@suse.cz>
13808 * doc/extend.texi: Document when a resolver function is
13809 generated for target_clones.
13811 2017-07-03 Martin Liska <mliska@suse.cz>
13813 * asan.c (asan_emit_stack_protection): Unpoison just red zones
13814 and shadow memory of auto variables which are subject of
13815 use-after-scope sanitization.
13816 (asan_expand_mark_ifn): Add do set only when is_poison.
13818 2016-07-03 Richard Biener <rguenther@suse.de>
13820 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
13822 (vect_force_simple_reduction): Record reduction def -> phi mapping.
13823 (vectorizable_reduction): Perform reduction PHI creation when
13824 visiting a reduction PHI and adjust and simplify code generation
13825 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
13826 (vect_transform_loop): Visit reduction PHIs.
13827 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
13828 defs into the SLP tree.
13829 (vect_build_slp_tree): Reduction defs terminate the recursion.
13830 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
13832 (vect_get_vec_defs_for_stmt_copy): Export.
13833 (vect_get_vec_defs): Likewise.
13834 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
13836 (vect_get_vec_defs_for_stmt_copy): Declare.
13837 (vect_get_vec_defs): Likewise.
13839 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13841 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
13842 parameter with a "loop" parameter and use it instead of the
13843 loop containing DR_STMT. Don't check simple_iv when doing
13844 BB analysis. Describe the two analysis modes in the comment.
13846 2017-07-03 Tom de Vries <tom@codesourcery.com>
13848 PR tree-optimization/69468
13849 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
13850 (find_same_succ_bb): Handle ignore_edge_flags.
13852 2017-07-03 Tom de Vries <tom@codesourcery.com>
13854 PR tree-optimization/81192
13855 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
13857 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
13859 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
13861 2017-07-03 Tom de Vries <tom@codesourcery.com>
13863 PR tree-optimization/81192
13864 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
13865 BB_SAME_SUCC (bb) == NULL.
13867 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13869 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
13872 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13874 * dumpfile.c: Include profile-count.h
13875 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
13877 (insert_cond_bb): Update profile.
13878 * tree-cfg.h (insert_cond_bb): Update prototype.
13879 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
13880 * tree-dump.c: Do not include tree-cfg.
13882 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13884 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
13886 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13888 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
13891 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13893 * tree-complex.c (expand_complex_div_wide): update profile.
13895 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13896 Alan Hayward <alan.hayward@arm.com>
13897 David Sherwood <david.sherwood@arm.com>
13899 * Makefile.in (MACHMODE_H): Remove insn-modes.h
13900 (CORETYPES_H): New define.
13901 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
13902 (insn-modes-inline.h, s-modes-inline-h): New rules.
13903 (generated_files): Add insn-modes-inline.h.
13904 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
13905 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
13906 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
13907 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
13908 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
13909 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
13910 (build/gencodes.o, build/genconditions.o): Likewise.
13911 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
13912 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
13913 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
13914 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
13915 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
13916 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
13917 * coretypes.h: Include everything up to real.h for generators.
13918 Include insn-modes.h first. Include wide-int-print.h after
13919 wide-int.h. Include insn-modes-inline.h and then machmode.h.
13920 * machmode.h: Don't include insn-modes.h here.
13921 * function-tests.c: Remove includes of signop.h, machmode.h,
13922 double-int.h and wide-int.h.
13924 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
13926 * optc-save-gen.awk: Likewise.
13927 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
13928 * godump.c: Remove include of wide-int-print.h.
13929 * pretty-print.h: Likewise.
13930 * wide-int-print.cc: Likewise.
13931 * wide-int.cc: Likewise.
13932 * hash-map-tests.c: Remove include of signop.h.
13933 * hash-set-tests.c: Likewise.
13934 * rtl-tests.c: Likewise.
13935 * mkconfig.sh: Remove include of machmode.h.
13936 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
13938 (emit_insn_modes_inline_h): ...this new function. Emit the code
13939 into an insn-modes-inline.h header file, adding appropriate
13940 include guards and end comments.
13941 (emit_insn_modes_c_header): Remove include of machmode.h.
13942 (emit_min_insn_modes_c_header): Include coretypes.h rather than
13944 (main): Handle -i flag and call emit_insn_modes_inline_h when
13947 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13949 * tree-ssa-strlen.c (strinfo): Rename the length field to
13950 nonzero_chars. Add a full_string_p field.
13951 (compare_nonzero_chars, zero_length_string_p): New functions.
13952 (get_addr_stridx): Add an offset_out parameter.
13953 Use compare_nonzero_chars.
13954 (get_stridx): Update accordingly. Use compare_nonzero_chars.
13955 (new_strinfo): Update after above changes to strinfo.
13956 (set_endptr_and_length): Set full_string_p.
13957 (get_string_length): Update after above changes to strinfo.
13958 (unshare_strinfo): Update call to new_strinfo.
13959 (maybe_invalidate): Likewise.
13960 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
13961 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
13962 as a uhwi instead of an shwi. Update after above changes to
13963 strinfo and new_strinfo.
13964 (zero_length_string): Assert that chainsi contains full strings.
13965 Use zero_length_string_p. Update call to new_strinfo.
13966 (adjust_related_strinfos): Update after above changes to strinfo.
13967 Copy full_string_p from origsi.
13968 (adjust_last_stmt): Use zero_length_string_p.
13969 (handle_builtin_strlen): Update after above changes to strinfo and
13970 new_strinfo. Install the lhs as the string length if the previous
13971 entry didn't describe a full string.
13972 (handle_builtin_strchr): Update after above changes to strinfo
13974 (handle_builtin_strcpy): Likewise.
13975 (handle_builtin_strcat): Likewise.
13976 (handle_builtin_malloc): Likewise.
13977 (handle_pointer_plus): Likewise.
13978 (handle_builtin_memcpy): Likewise. Track nonzero characters
13979 that aren't necessarily followed by a nul terminator.
13980 (handle_char_store): Likewise.
13982 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13984 PR tree-optimization/80769
13985 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
13986 for malloc and calloc. Document the new invariant that all related
13987 strinfos have delayed lengths or none do.
13988 (verify_related_strinfos): Move earlier in file.
13989 (set_endptr_and_length): New function, split out from...
13990 (get_string_length): ...here. Also set the lengths of related
13992 (zero_length_string): Assert that chainsi has known (rather than
13994 (adjust_related_strinfos): Likewise.
13996 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13998 PR tree-optimization/81136
13999 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
14000 assert that two references with the same misalignment have the same
14001 compile-time misalignment if those compile-time misalignments
14004 2017-07-01 Andi Kleen <ak@linux.intel.com>
14006 * print-tree.c (print_node): Print all attributes.
14008 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14010 * cfg.c (scale_bbs_frequencies): New function.
14011 * cfg.h (scale_bbs_frequencies): Declare it.
14012 * cfgloopanal.c (single_likely_exit): Cleanup.
14013 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
14015 (scale_loop_profile): Likewise.
14016 (loop_version): Likewise.
14017 (create_empty_loop_on_edge): Update.
14018 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
14019 scale_loop_frequencies, scale_loop_profile, loopify,
14020 loop_version): Update prototypes.
14021 * modulo-sched.c (sms_schedule): Update.
14022 * predict.c (unlikely_executed_edge_p): Also check probability.
14023 (probably_never_executed_edge_p): Fix typo.
14024 * tree-if-conv.c (version_loop_for_if_conversion): Update.
14025 * tree-parloops.c (gen_parallel_loop): Update.
14026 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
14027 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14028 * tree-ssa-loop-split.c (split_loop): Update.
14029 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
14030 * tree-vect-loop-manip.c (vect_do_peeling): Update.
14031 (vect_loop_versioning): Update.
14032 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14034 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14036 * trans-mem.c (split_bb_make_tm_edge): Update profile.
14038 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14040 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
14041 to keep profile consistent.
14043 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14045 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
14046 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
14047 * profile-count.h (max_safe_multiplier): Make unsigned.
14048 (profile_count::guessed_zero): New.
14050 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14052 * bb-reorder.c (fix_up_crossing_landing_pad,
14053 fix_crossing_conditional_branches): Use make_single_succ_edge
14054 to keep profile consistent.
14056 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14058 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
14061 2017-07-01 Jakub Jelinek <jakub@redhat.com>
14064 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
14065 the right scopes, make sure cond_jump isn't preserved between multiple
14066 iterations. Search for fallthru edge whenever there are 3+ edges and
14067 use find_fallthru_edge for it.
14069 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14071 Patch by Alexander Monakov <amonakov@ispras.ru>
14072 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
14073 probabilities consistently.
14075 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14077 * pa.c (pa_expand_compare_and_swap_loop): Update call of
14078 emit_cmp_and_jump_insns.
14080 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14083 * tree-inline.c (expand_call_inline): Combine profile statuses.
14085 2017-06-30 Andrew Pinski <apinski@cavium.com>
14087 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
14088 fold_stmt returned true.
14090 2017-06-30 Nathan Sidwell <nathan@acm.org>
14092 * ggc.h (empty_string): Delete.
14093 * cfgexpand.c (expand_asm_stmt): Use plain "".
14094 * optabs.c (expand_asm_memory_barrier): Likewise.
14095 * stringpool.c (empty_string): Delete.
14096 (digit_vector, digit_string): Delete.
14097 (ggc_alloc_string): Use plain "", don't optimize single digit
14098 strings. Use ggc_alloc_atomic.
14100 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
14102 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
14103 comparison set and one other set, use the cost of the non-comparison
14106 2017-06-30 Nathan Sidwell <nathan@acm.org>
14108 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
14111 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
14113 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
14114 loops. Remove now unneeded calls to gimple_switch_set_label() that
14115 just set removed labels to NULL_TREE.
14117 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
14119 * tree-ssanames.c (set_range_info_raw): Abstract from ...
14120 (set_range_info): ...here. Only call set_range_info_raw if domain
14122 (set_nonzero_bits): Call set_range_info_raw.
14123 * tree-ssanames.h (set_range_info_raw): New.
14125 2017-06-30 Jakub Jelinek <jakub@redhat.com>
14128 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
14129 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
14130 of nonimmediate_operand and <store_mask_constraint> instead of m
14131 for the input operand. For V8FI iterator, always split if input
14132 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
14133 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
14134 <store_mask_predicate> instead of register_operand and
14135 <store_mask_constraint> instead of v for the input operand. Make
14136 sure both operands aren't MEMs for if not <mask_applied>.
14138 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
14140 * lto-wrapper.c (copy_file) Close both file descriptors before
14143 2017-06-30 Martin Liska <mliska@suse.cz>
14146 * multiple_target.c (create_dispatcher_calls): Make ifunc
14147 also for function that don't have calls or are not referenced.
14149 2017-06-30 Richard Biener <rguenther@suse.de>
14151 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
14152 analyze the first scalar stmt. Move vector type computation
14153 for the BB case here from ...
14154 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
14155 live operation processing in the SLP case properly.
14157 2017-06-30 Richard Biener <rguenther@suse.de>
14159 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
14161 2017-06-30 Martin Liska <mliska@suse.cz>
14164 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
14165 before BUILT_IN_UNWIND_RESUME when ASAN is used.
14167 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
14169 * doc/invoke.texi (AArch64): Add missing options and remove redundant
14172 2017-06-30 Richard Biener <rguenther@suse.de>
14174 PR tree-optimization/81249
14175 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
14176 condition reduction result to original scalar type.
14178 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14180 * profile-count.h (enum profile_quality): Fix typos and whitespace
14183 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14185 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
14186 type for branch probabilities.
14188 2017-06-29 Julian Brown <julian@codesourcery.com>
14189 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14191 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
14192 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
14193 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
14194 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
14196 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14198 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
14199 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
14200 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
14201 CC usage from generic code to here.
14202 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
14203 CC usage into the target macros.
14205 2017-06-29 Maya Rashish <coypu@sdf.org>
14207 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
14210 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14212 * arm/arm-builtins.c: Include profile-count.h
14213 * except.c (sjlj_emit_function_enter): Use
14214 profile_probability::unlikely.
14216 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14218 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
14219 and tocrel_offset be pointer args rather than implicitly using
14221 (legitimate_constant_pool_address_p, rs6000_emit_move,
14222 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
14223 tocrel_offset and use in toc_relative_expr_p call.
14224 (print_operand, print_operand_address): Use static tocrel_base_oac
14225 and tocrel_offset_oac.
14226 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
14229 2017-06-29 Maya Rashish <coypu@sdf.org>
14231 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
14233 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
14235 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
14236 objects, take into account only the alignment of 'op0' and 'mode1' if
14239 2017-06-29 Steve Ellcey <sellcey@cavium.com>
14241 * ccmp.c (ccmp_tree_comparison_p): New function.
14242 (ccmp_candidate_p): Update to use above function.
14243 (get_compare_parts): New function.
14244 (expand_ccmp_next): Update to use new functions.
14245 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
14247 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
14248 take mode as argument.
14249 * ccmp.h (expand_ccmp_expr): Add mode as argument.
14250 * expr.c (expand_expr_real_1): Pass mode as argument.
14252 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
14254 * combine.c (combine_instructions): Print insns to dump_file, together
14257 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14259 * asan.c (asan_emit_stack_protection): Update.
14260 (create_cond_insert_point): Update.
14261 * auto-profile.c (afdo_propagate_circuit): Update.
14262 * basic-block.h (struct edge_def): Turn probability to
14263 profile_probability.
14264 (EDGE_FREQUENCY): Update.
14265 * bb-reorder.c (find_traces_1_round): Update.
14266 (better_edge_p): Update.
14267 (sanitize_hot_paths): Update.
14268 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
14269 (make_single_succ_edge): Update.
14270 (check_bb_profile): Update.
14271 (dump_edge_info): Update.
14272 (update_bb_profile_for_threading): Update.
14273 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
14274 probabilitycount to 0.
14275 * cfgbuild.c (compute_outgoing_frequencies): Update.
14276 * cfgcleanup.c (try_forward_edges): Update.
14277 (outgoing_edges_match): Update.
14278 (try_crossjump_to_edge): Update.
14279 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
14280 (expand_gimple_tailcall): Update.
14281 (construct_init_block): Use make_single_succ_edge.
14282 (construct_exit_block): Use make_single_succ_edge.
14283 * cfghooks.c (verify_flow_info): Update.
14284 (redirect_edge_succ_nodup): Update.
14285 (split_edge): Update.
14286 (account_profile_record): Update.
14287 * cfgloopanal.c (single_likely_exit): Update.
14288 * cfgloopmanip.c (scale_loop_profile): Update.
14289 (set_zero_probability): Remove.
14290 (duplicate_loop_to_header_edge): Update.
14291 * cfgloopmanip.h (loop_version): Update prototype.
14292 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
14293 (force_nonfallthru_and_redirect): Update.
14294 (update_br_prob_note): Update.
14295 (rtl_verify_edges): Update.
14296 (purge_dead_edges): Update.
14297 (rtl_lv_add_condition_to_bb): Update.
14298 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
14299 * cgraphunit.c (init_lowered_empty_function): Update.
14300 (cgraph_node::expand_thunk): Update.
14301 * cilk-common.c: Include profile-count.h
14302 * dojump.c (inv): Remove.
14303 (jumpifnot): Update.
14304 (jumpifnot_1): Update.
14305 (do_jump_1): Update.
14307 (do_jump_by_parts_greater_rtx): Update.
14308 (do_compare_rtx_and_jump): Update.
14309 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
14310 do_jump_1. do_compare_rtx_and_jump): Update prototype.
14311 * dwarf2cfi.c: Include profile-count.h
14312 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
14313 (sjlj_emit_dispatch_table): Likewise.
14314 * explow.c: Include profile-count.h
14315 * expmed.c (emit_store_flag_force): Update.
14316 (do_cmp_and_jump): Update.
14317 * expr.c (compare_by_pieces_d::generate): Update.
14318 (compare_by_pieces_d::finish_mode): Update.
14319 (emit_block_move_via_loop): Update.
14320 (store_expr_with_bounds): Update.
14321 (store_constructor): Update.
14322 (expand_expr_real_2): Update.
14323 (expand_expr_real_1): Update.
14324 * expr.h (try_casesi, try_tablejump): Update prototypes.
14325 * gimple-pretty-print.c (dump_probability): Update.
14326 (dump_profile): New.
14327 (dump_gimple_label): Update.
14328 (dump_gimple_bb_header): Update.
14329 * graph.c (draw_cfg_node_succ_edges): Update.
14330 * hsa-gen.c (convert_switch_statements): Update.
14331 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
14332 (find_if_case_1): Update.
14333 (find_if_case_2): Update.
14334 * internal-fn.c (expand_arith_overflow_result_store): Update.
14335 (expand_addsub_overflow): Update.
14336 (expand_neg_overflow): Update.
14337 (expand_mul_overflow): Update.
14338 (expand_vector_ubsan_overflow): Update.
14339 * ipa-cp.c (good_cloning_opportunity_p): Update.
14340 * ipa-split.c (split_function): Use make_single_succ_edge.
14341 * ipa-utils.c (ipa_merge_profiles): Update.
14342 * loop-doloop.c (add_test): Update.
14343 (doloop_modify): Update.
14344 * loop-unroll.c (compare_and_jump_seq): Update.
14345 (unroll_loop_runtime_iterations): Update.
14346 * lra-constraints.c (lra_inheritance): Update.
14347 * lto-streamer-in.c (input_cfg): Update.
14348 * lto-streamer-out.c (output_cfg): Update.
14349 * mcf.c (adjust_cfg_counts): Update.
14350 * modulo-sched.c (sms_schedule): Update.
14351 * omp-expand.c (expand_omp_for_init_counts): Update.
14352 (extract_omp_for_update_vars): Update.
14353 (expand_omp_ordered_sink): Update.
14354 (expand_omp_for_ordered_loops): Update.
14355 (expand_omp_for_generic): Update.
14356 (expand_omp_for_static_nochunk): Update.
14357 (expand_omp_for_static_chunk): Update.
14358 (expand_cilk_for): Update.
14359 (expand_omp_simd): Update.
14360 (expand_omp_taskloop_for_outer): Update.
14361 (expand_omp_taskloop_for_inner): Update.
14362 * omp-simd-clone.c (simd_clone_adjust): Update.
14363 * optabs.c (expand_doubleword_shift): Update.
14364 (expand_abs): Update.
14365 (emit_cmp_and_jump_insn_1): Update.
14366 (expand_compare_and_swap_loop): Update.
14367 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
14368 * predict.c (predictable_edge_p): Update.
14369 (edge_probability_reliable_p): Update.
14370 (set_even_probabilities): Update.
14371 (combine_predictions_for_insn): Update.
14372 (combine_predictions_for_bb): Update.
14373 (propagate_freq): Update.
14374 (estimate_bb_frequencies): Update.
14375 (force_edge_cold): Update.
14376 * profile-count.c (profile_count::dump): Add missing space into dump.
14377 (profile_count::debug): Add newline.
14378 (profile_count::differs_from_p): Explicitly convert to unsigned.
14379 (profile_count::stream_in): Update.
14380 (profile_probability::dump): New member function.
14381 (profile_probability::debug): New member function.
14382 (profile_probability::differs_from_p): New member function.
14383 (profile_probability::differs_lot_from_p): New member function.
14384 (profile_probability::stream_in): New member function.
14385 (profile_probability::stream_out): New member function.
14386 * profile-count.h (profile_count_quality): Rename to ...
14387 (profile_quality): ... this one.
14388 (profile_probability): New.
14389 (profile_count): Update.
14390 * profile.c (compute_branch_probabilities): Update.
14391 * recog.c (peep2_attempt): Update.
14392 * sched-ebb.c (schedule_ebbs): Update.
14393 * sched-rgn.c (find_single_block_region): Update.
14394 (compute_dom_prob_ps): Update.
14395 (schedule_region): Update.
14396 * sel-sched-ir.c (compute_succs_info): Update.
14397 * stmt.c (struct case_node): Update.
14398 (do_jump_if_equal): Update.
14399 (get_outgoing_edge_probs): Update.
14400 (conditional_probability): Update.
14401 (emit_case_dispatch_table): Update.
14402 (expand_case): Update.
14403 (expand_sjlj_dispatch_table): Update.
14404 (emit_case_nodes): Update.
14405 * targhooks.c: Update.
14406 * tracer.c (better_p): Update.
14407 (find_best_successor): Update.
14408 * trans-mem.c (expand_transaction): Update.
14409 * tree-call-cdce.c: Update.
14410 * tree-cfg.c (gimple_split_edge): Upate.
14411 (move_sese_region_to_fn): Upate.
14412 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
14413 * tree-eh.c (lower_resx): Upate.
14414 (cleanup_empty_eh_move_lp): Upate.
14415 * tree-if-conv.c (version_loop_for_if_conversion): Update.
14416 * tree-inline.c (copy_edges_for_bb): Update.
14417 (copy_cfg_body): Update.
14418 * tree-parloops.c (gen_parallel_loop): Update.
14419 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
14420 (gimple_gen_time_profiler): Update.
14421 * tree-ssa-dce.c (remove_dead_stmt): Update.
14422 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
14423 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
14424 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
14425 (unloop_loops): Update.
14426 (try_peel_loop): Update.
14427 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14428 * tree-ssa-loop-split.c (connect_loops): Update.
14429 (split_loop): Update.
14430 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
14431 (hoist_guard): Update.
14432 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
14433 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
14434 (value_replacement): Update.
14435 * tree-ssa-reassoc.c (branch_fixup): Update.
14436 * tree-ssa-tail-merge.c (replace_block_by): Update.
14437 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
14438 (create_edge_and_update_destination_phis): Update.
14439 (compute_path_counts): Update.
14440 (recompute_probabilities): Update.
14441 (update_joiner_offpath_counts): Update.
14442 (freqs_to_counts_path): Update.
14443 (duplicate_thread_path): Update.
14444 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
14445 (struct switch_conv_info): Update.
14446 (gen_inbound_check): Update.
14447 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
14448 (vect_do_peeling): Update.
14449 (vect_loop_versioning): Update.
14450 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14451 (optimize_mask_stores): Update.
14452 * ubsan.c (ubsan_expand_null_ifn): Update.
14453 * value-prof.c (gimple_divmod_fixed_value): Update.
14454 (gimple_divmod_fixed_value_transform): Update.
14455 (gimple_mod_pow2): Update.
14456 (gimple_mod_pow2_value_transform): Update.
14457 (gimple_mod_subtract): Update.
14458 (gimple_mod_subtract_transform): Update.
14459 (gimple_ic): Update.
14460 (gimple_stringop_fixed_value): Update.
14461 (gimple_stringops_transform): Update.
14462 * value-prof.h: Update.
14464 2017-06-29 Carl Love <cel@us.ibm.com>
14466 * config/rs6000/rs6000-c.c: Add support for built-in functions
14467 vector signed int vec_signed (vector float);
14468 vector signed long long vec_signed (vector double);
14469 vector signed int vec_signed2 (vector double, vector double);
14470 vector signed int vec_signede (vector double);
14471 vector signed int vec_signedo (vector double);
14472 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
14473 instruction generator.
14474 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
14475 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
14476 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
14478 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
14479 vunsignede_v2df): Add define_expands.
14480 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
14481 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
14482 VEC_UNSIGNEDO): Add definitions.
14483 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
14484 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
14485 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
14486 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
14487 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
14488 * config/rs6000/altivec.h (vec_signed, vec_signed2,
14489 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
14490 vec_unsignede, vec_unsignedo): Add builtin defines.
14491 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
14493 * doc/extend.texi: Update the built-in documentation file for the
14494 new built-in functions.
14496 2017-06-29 Richard Biener <rguenther@suse.de>
14498 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
14499 reduction chains to LOOP_VINFO_REDUCTIONS.
14500 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
14501 SLP reductions after processing reduction chains.
14503 2017-06-29 Nathan Sidwell <nathan@acm.org>
14505 * builtins.c (fold_builtin_FUNCTION): Use
14506 lang_hooks.decl_printable_name.
14508 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
14510 PR middle-end/81194
14511 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
14512 with only one label.
14513 * stmt.c (expand_case): Assert NCASES is greater than one.
14515 2017-06-29 Richard Biener <rguenther@suse.de>
14517 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
14519 (group_case_labels): Likewise.
14520 (find_taken_edge): Push sanity checking on val to workers...
14521 (find_taken_edge_cond_expr): ... here
14522 (find_taken_edge_switch_expr): ... and here, handle cases
14523 with just a default label.
14524 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
14525 (group_case_labels): Likewise.
14526 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
14527 group_case_labels does anything cleanup the CFG again.
14529 2017-06-29 Bin Cheng <bin.cheng@arm.com>
14531 PR tree-optimization/81196
14532 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
14533 exit condition comparing two IVs.
14535 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
14537 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
14538 profile to the dummy entry at the end of the list of architectures.
14539 * config/arm/arm-cpu-cdata.h: Regenerated.
14541 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14542 Michael Collison <michael.collison@arm.com>
14545 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
14547 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
14548 (*aarch64_reg_<mode>3_minus_mask): New pattern.
14549 (*aarch64_<optab>_reg_di3_mask2): New pattern.
14550 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
14551 of shift when the shift amount is masked with constant equal to
14552 the size of the mode.
14553 * config/aarch64/predicates.md (subreg_lowpart_operator): New
14556 2017-06-29 Martin Liska <mliska@suse.cz>
14558 * config/i386/i386.opt: Change range from [1,5] to [0,5].
14560 2017-06-29 Yury Gribov <tetra2005@gmail.com>
14563 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
14566 (edge_growth_cache_entry::edge_growth_cache_entry): New
14568 (reset_edge_growth_cache): Update to use constructor.
14570 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14572 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
14573 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
14574 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
14576 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
14578 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
14579 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
14581 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
14583 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
14584 (*-linux-uclibc*): Add t-uclibc tmake_file.
14585 * config/t-musl: New.
14586 * config/t-uclibc: New.
14588 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
14590 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
14592 (gen_comm_data): Emit architectural setting of arch_prof.
14593 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
14595 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
14596 (armv8-m.base, armv8-m.main): Likewise.
14597 * arm-protos.h (arm_build_target): Add profile field.
14598 (arch_option): Likewise.
14599 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
14601 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
14602 arm_active_target.profile.
14604 2017-06-28 Richard Biener <rguenther@suse.de>
14606 PR middle-end/81227
14607 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
14608 TYPE_OVERFLOW_WRAPS.
14609 * match.pd (negate_expr_p): Likewise.
14610 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
14611 fold_build2, not fold_binary.
14613 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
14615 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
14616 Convert memory address to Pmode.
14617 (aarch64_print_operand): Assert MEM operands are always Pmode.
14619 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
14622 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
14623 Remove redundant if.
14624 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
14625 * config/arm/aarch-common-protos.h
14626 (aarch_forward_to_shift_is_not_shifted_re): Remove.
14627 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
14629 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
14632 * multiple_target.c (create_dispatcher_calls): Set the default
14633 clone to be static, not public.
14635 2017-06-28 Richard Biener <rguenther@suse.de>
14637 * tree-vect-loop.c (vectorizable_reduction): Move special
14638 cond reduction IV var creation ...
14639 (vect_create_epilog_for_reduction): ... here. Remove induction_index
14640 parameter. Use STMT_VINFO_VECTYPE.
14641 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
14644 2017-06-28 Martin Liska <mliska@suse.cz>
14647 * ipa-visibility.c (non_local_p): Handle visibility.
14649 2017-06-28 Martin Liska <mliska@suse.cz>
14652 * common.opt: Add IntegerRange to various options.
14653 * opt-functions.awk (integer_range_info): New function.
14654 * optc-gen.awk: Add integer_range_info to cl_options struct.
14655 * opts-common.c (decode_cmdline_option): Handle
14656 CL_ERR_INT_RANGE_ARG.
14657 (cmdline_handle_error): Likewise.
14658 * opts.c (print_filtered_help): Show valid interval in
14659 when --help is provided.
14660 * opts.h (struct cl_option): Add range_min and range_max fields.
14661 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
14663 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
14665 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
14666 (x * C EQ/NE y * C): New transformation.
14668 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
14670 * genmultilib (combination_space): Accept '+' in option names.
14672 2017-06-28 Martin Liska <mliska@suse.cz>
14675 * asan.c (instrument_derefs): Bail out inner references
14676 that are hard register variables.
14678 2017-06-28 Jakub Jelinek <jakub@redhat.com>
14681 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
14682 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
14684 2017-06-28 Richard Biener <rguenther@suse.de>
14686 * tree-vectorizer.h (vect_get_vec_defs): Remove.
14687 (vect_get_slp_defs): Adjust.
14688 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
14690 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
14692 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
14693 get_initial_defs_for_reduction instead of vect_get_vec_defs.
14694 (vectorizable_reduction): Adjust.
14695 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
14697 (vect_get_slp_defs): Likewise.
14698 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
14699 (vectorizable_bswap): Adjust.
14700 (vectorizable_call): Likewise.
14701 (vectorizable_conversion): Likewise.
14702 (vectorizable_assignment): Likewise.
14703 (vectorizable_shift): Likewise.
14704 (vectorizable_operation): Likewise.
14705 (vectorizable_store): Likewise.
14706 (vectorizable_condition): Likewise.
14707 (vectorizable_comparison): Likewise.
14709 2017-06-28 Michael Collison <michael.collison@arm.com>
14712 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
14714 (arm_gen_movmemqi): Removed unused variable 'i'.
14715 Convert 'for' loop into 'while' loop.
14716 (arm_expand_prologue): Remove last unnecessary set of insn.
14717 (thumb_pop): Remove unused variable 'pushed_words'.
14718 (thumb_exit): Remove last unnecessary set of regs_to_pop.
14720 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14722 * config/s390/predicates.md: Use s390_rel_address_ok_p.
14723 * config/s390/s390-protos.h: Add prototype of
14724 s390_rel_address_ok_p.
14725 * config/s390/s390.c (s390_got_symbol): New function.
14726 (s390_rel_address_ok_p): New function.
14727 (legitimize_pic_address): Use s390_rel_address_ok_p.
14728 (s390_load_got): Use s390_got_symbol.
14729 (s390_option_override): Issue error if
14730 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
14731 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
14733 * config/s390/s390.opt: New option mpic-data-is-text-relative.
14735 2017-06-27 Andrew Pinski <apinski@cavium.com>
14737 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
14738 (X * copysign (1.0, X)): New pattern.
14739 (X * copysign (1.0, -X)): New pattern.
14740 (copysign (-1.0, CST)): New pattern.
14742 2017-06-27 Joseph Myers <joseph@codesourcery.com>
14744 * genmultilib (combination_space): Remove variable.
14745 Validate reuse rules against regular expression for any sequence
14746 of multilib options in any order.
14748 2017-06-27 Michael Collison <michael.collison@arm.com>
14750 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
14751 call aarch64_split_simd_combine.
14752 * (aarch64_combine_internal<mode>): Delete pattern.
14753 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
14754 Allow register and subreg operands.
14756 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14758 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
14759 specific need, just fallback on defaults.
14760 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
14762 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14763 Olivier Hainque <hainque@adacore.com>
14765 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
14767 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
14768 targets. Pick a default if no particular attempt applied.
14769 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
14772 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14774 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
14775 (x86_64-wrs-vxworks7): Likewise.
14777 2017-06-27 Marek Polacek <polacek@redhat.com>
14780 * ubsan.c (instrument_null): Check get_base_address's result for null.
14782 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
14784 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
14786 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
14788 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
14789 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
14790 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
14791 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
14792 New function types.
14793 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
14794 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
14795 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
14796 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
14797 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
14798 BUILT_IN_FEUPDATEENV): New builtins.
14799 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
14800 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
14801 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
14802 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
14804 (builtin_structptr_types): Adjust size.
14805 * tree.c (builtin_structptr_types): Add four entries.
14807 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14808 Olivier Hainque <hainque@adacore.com>
14810 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
14811 (TLS_SYM): New local macro, forcing reference to __tls__ on
14812 link command lines for VxWorks 7 RTPs, triggering initialization
14814 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
14815 OS features TLS support, true for RTPs on VxWorks 7.
14816 * config/vxworks.c (vxworks_override_options): Setup emutls
14819 2017-06-27 Jakub Jelinek <jakub@redhat.com>
14821 * predict.c (test_prediction_value_range): Use -1U instead of -1
14822 to avoid narrowing conversion warning.
14823 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
14824 to avoid narrowing conversion warning.
14825 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
14827 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
14829 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14831 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
14832 64bit configurations.
14833 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
14834 (SIZE_TYPE): Likewise.
14835 * config/vxworks.c (vxworks_emutls_var_fields): Use
14836 long_unsigned_type_node instead of unsigned_type_node as the offset
14837 field type, which is "pointer" mode in emutls.c.
14839 2017-06-27 Jakub Jelinek <jakub@redhat.com>
14842 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
14844 PR middle-end/81207
14845 * gimple-fold.c (replace_call_with_call_and_fold): Handle
14846 gimple_vuse copying separately from gimple_vdef copying.
14848 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14850 * value-prof.c (free_hist): Remove call to memset and the enclosing if
14853 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
14854 Olivier Hainque <hainque@adacore.com>
14856 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
14857 for all vxworks7 targets.
14858 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
14859 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
14860 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
14861 variations for VX6/VX7 and 32/64bits later on in ...
14862 (VXWORKS_LIB_SPEC): Leverage new macros.
14863 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
14864 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
14866 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
14868 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
14869 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
14871 2017-06-26 Carl Love <cel@us.ibm.com>
14873 * config/rs6000/rs6000-c.c: Add support for built-in functions
14874 vector bool char vec_reve (vector bool char);
14875 vector signed char vec_reve (vector signed char);
14876 vector unsigned char vec_reve (vector unsigned char);
14877 vector bool int vec_reve (vector bool int);
14878 vector signed int vec_reve (vector signed int);
14879 vector unsigned int vec_reve (vector unsigned int);
14880 vector bool long long vec_reve (vector bool long long);
14881 vector signed long long vec_reve (vector signed long long);
14882 vector unsigned long long vec_reve (vector unsigned long long);
14883 vector bool short vec_reve (vector bool short);
14884 vector signed short vec_reve (vector signed short);
14885 vector double vec_reve (vector double);
14886 vector float vec_reve (vector float);
14887 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
14888 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
14889 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
14890 (altivec_vreve): New pattern.
14891 * config/rs6000/altivec.h (vec_reve): New define.
14892 * doc/extend.texi (vec_rev): Update the built-in documentation file
14893 for the new built-in functions.
14895 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14897 PR tree-optimization/71815
14898 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
14900 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
14902 (slsr_process_phi): Likewise.
14903 (replace_uncond_cands_and_profitable_phis): Don't replace a
14904 multiply candidate with a stride of 1 (copy or cast).
14905 (phi_incr_cost): Call uses_consumed_by_stmt rather than
14907 (lowest_cost_path): Likewise.
14908 (total_savings): Likewise.
14910 2017-06-26 Richard Biener <rguenther@suse.de>
14913 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
14914 Use def_builtin_pure for all gather builtins.
14916 2017-06-26 Richard Biener <rguenther@suse.de>
14918 PR tree-optimization/81203
14919 * tree-tailcall.c (find_tail_calls): Do not move stmts into
14920 non-dominating BBs.
14922 2017-06-26 Marek Polacek <polacek@redhat.com>
14925 * doc/invoke.texi: Document -Wmultistatement-macros.
14927 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
14929 * doc/sourcebuild.texi (ARM-specific attributes): Document new
14930 arm_neon_ok_no_float_abi effective target.
14932 2017-06-26 Richard Biener <rguenther@suse.de>
14934 PR tree-optimization/80928
14935 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
14936 (copy_bbs): Set BB_DUPLICATED flag early.
14937 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
14939 (execute_on_shrinking_pred): Likewise.
14940 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
14941 BB_DUPLICATED blocks.
14942 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
14943 iterate over all PHIs considering removal of *gsi.
14945 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
14947 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
14950 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14952 * config/rs6000/rs6000-string.c: (expand_block_clear,
14953 do_load_for_compare, select_block_compare_mode,
14954 compute_current_alignment, expand_block_compare,
14955 expand_strncmp_align_check, expand_strn_compare,
14956 expand_block_move, rs6000_output_load_multiple)
14957 Move functions related to string/block move/compare
14958 to a separate file.
14959 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
14960 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
14961 for this function which is now used in two files.
14962 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
14963 * config.gcc: Add rs6000-string.o to extra_objs for
14964 targets powerpc*-*-* and rs6000*-*-*.
14966 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
14969 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
14970 32-bit, since indexed is not valid for DImode.
14971 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
14972 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
14973 (define_peephole2 for Altivec d-form load): Add 32-bit support.
14974 (define_peephole2 for Altivec d-form store): Likewise.
14977 * multiple_target.c (create_dispatcher_calls): Only create the
14978 dispatcher call if the function is the default clone of a
14979 versioned function.
14981 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
14983 PR middle-end/80902
14984 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
14985 a call, force the call to not be a tail call.
14987 2017-06-23 Jeff Law <law@redhat.com>
14989 * doc/contrib.texi: Add entry for Steven Pemberton's work on
14992 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
14994 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
14995 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
14996 handling for early expansion of vector shifts (sl,sr,sra,rl).
14997 (builtin_function_type): Add vector shift right instructions
14998 to the unsigned argument list.
15000 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
15002 rtl-optimizatoin/79286
15003 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
15004 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
15005 trap. PIC register plus a const unspec without offset can never trap.
15007 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
15009 * tree.h (builtin_structptr_type): New type.
15010 (builtin_structptr_types): Declare new array.
15011 * tree.c (builtin_structptr_types): New array.
15012 (free_lang_data, build_common_tree_nodes): Use it.
15014 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
15017 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
15020 2017-06-22 Matt Turner <mattst88@gmail.com>
15022 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
15023 Lake models to skylake case. Assume skylake for unknown
15024 models with clflushopt.
15026 2017-06-22 Jeff Law <law@redhat.com>
15028 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
15029 frame sizes that do not satisfy aarch64_uimm12_shift.
15031 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
15033 * profile-count.h (apply_probability,
15034 apply_scale, probability_in): Fix checks for zero.
15036 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15038 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
15039 * doc/cppdiropts.texi (-I @var{dir}): Document it.
15041 2016-06-22 Richard Biener <rguenther@suse.de>
15043 * tree-vect-loop.c (vect_model_reduction_cost): Handle
15044 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
15045 REDUC_MAX_EXPR support.
15046 (vectorizable_reduction): Likewise.
15047 (vect_create_epilog_for_reduction): Likewise.
15049 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
15051 * match.pd (A / (1 << B) -> A >> B): New.
15052 * generic-match-head.c: Include optabs-tree.h.
15053 * gimple-match-head.c: Likewise.
15054 * optabs-tree.h (target_supports_op_p): New.
15055 * optabs-tree.c (target_supports_op_p): New.
15057 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15059 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
15060 $gcc_cv_ld --help output.
15061 (gcc_cv_ld_demangle): Likewise.
15062 (gcc_cv_ld_eh_frame_hdr): Likewise.
15063 (gcc_cv_ld_pie): Likewise.
15064 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
15065 (gcc_cv_ld_buildid): Likewise.
15066 (gcc_cv_ld_sysroot): Likewise.
15067 (ld_bndplt_support): Likewise.
15068 (ld_pushpopstate_support): Likewise.
15069 * configure: Regenerate.
15070 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
15072 2017-06-21 Jakub Jelinek <jakub@redhat.com>
15075 * config/i386/sse.md (round<mode>2): Renumber match_dup and
15076 operands indexes to avoid gap between operands and match_dups.
15078 2017-06-21 Andrew Pinski <apinski@cavium.com>
15080 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
15081 Increment Arith_shift and Arith_shift_reg by 1.
15082 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
15084 * config/aarch64/aarch64.c (thunderx_tunings): Enable
15085 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
15086 (aarch64_strip_extend): Add new argument and test for it.
15087 (aarch64_cheap_mult_shift_p): New function.
15088 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
15089 add a cost if it is true.
15090 Update calls to aarch64_strip_extend.
15091 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
15093 2017-06-21 Andrew Pinski <apinski@cavium.com>
15095 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
15097 (thunderxt88): Likewise.
15098 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
15099 (thunderx_prefetch_tune): New variable.
15100 (thunderx2t99_prefetch_tune): Update for the correct values.
15101 (thunderxt88_tunings): New variable.
15102 (thunderx_tunings): Use thunderx_prefetch_tune instead of
15103 generic_prefetch_tune.
15104 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
15106 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15108 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
15109 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
15110 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
15111 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
15112 (aarch64_atomic_cas<mode>, GPI): Likewise.
15114 2017-06-21 Martin Liska <mliska@suse.cz>
15116 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
15117 statements on cold and hot labels.
15118 * predict.c (tree_estimate_probability_bb): Remove the
15119 prediction from this place.
15121 2017-06-21 Martin Liska <mliska@suse.cz>
15123 PR tree-optimization/79489
15124 * gimplify.c (maybe_add_early_return_predict_stmt): New
15126 (gimplify_return_expr): Call the function.
15127 * predict.c (tree_estimate_probability_bb): Remove handling
15129 * predict.def: Update comment about early return predictor.
15130 * gimple-predict.h (is_gimple_predict): New function.
15131 * predict.def: Change default value of early return to 66.
15132 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
15134 * passes.def: Put pass_strip_predict_hints to the beginning of
15137 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
15139 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
15140 FUNCTION_DECL declarations.
15141 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
15143 (dwaf2out_decl): Likewise.
15144 * godump.c (go_early_global_decl): Skip call to the real debug hook
15145 for FUNCTION_DECL declarations.
15146 * passes.c (rest_of_decl_compilation): Skip call to the
15147 early_global_decl debug hook for FUNCTION_DECL declarations, unless
15148 -fdump-go-spec is passed.
15150 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
15152 * config/i386/i386.c (struct builtin_isa): New field pure_p.
15153 Reorder for compactness.
15154 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
15155 (def_builtin_pure, def_builtin_pure2): New functions.
15156 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
15158 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
15160 * match.pd (nop_convert): New predicate.
15161 ((A +- CST1) +- CST2): Allow some NOP conversions.
15163 2017-06-21 Jakub Jelinek <jakub@redhat.com>
15166 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
15167 with ctors/dtors if GOVD_SHARED is set.
15169 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
15171 * config/aarch64/aarch64.md (movti_aarch64):
15172 Emit mov rather than orr.
15173 (movtf_aarch64): Likewise.
15174 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
15175 Emit mov rather than orr.
15177 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
15179 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
15180 Swap alternatives, make integer dup more expensive.
15182 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
15184 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
15185 Return true for non-tls symbols.
15187 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
15189 * config/aarch64/aarch64-cores.def (cortex-a55): New.
15190 (cortex-a75): Likewise.
15191 (cortex-a75.cortex-a55): Likewise.
15192 * config/aarch64/aarch64-tune.md: Regenerate.
15193 * doc/invoke.texi (-mtune): Document new values for -mtune.
15195 2017-06-21 Tom de Vries <tom@codesourcery.com>
15197 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
15198 stack_size feature.
15199 (Effective-Target Keywords, Other attributes): Suggest using
15200 dg-add-options stack_size feature to get stack limit in stack_size
15201 effective target documentation.
15203 2017-06-21 Julian Brown <julian@codesourcery.com>
15204 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15206 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
15207 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
15208 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
15210 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
15211 attribute type list for neon_multiply.
15212 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
15213 attribute type list for neon_multiply.
15214 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
15215 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
15216 attribute type list for neon_multiply.
15217 * config/arm/types.md (crypto_pmull): Add.
15218 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
15219 attribute type list.
15221 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
15223 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
15226 2017-06-20 Jakub Jelinek <jakub@redhat.com>
15228 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
15229 to make sure not to dereference a NULL cost_classes_ptr pointer.
15231 2017-06-20 Carl Love <cel@us.ibm.com>
15233 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15234 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15235 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15236 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15237 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15238 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15239 VMULOSW): New enum "unspec" values.
15240 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
15241 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
15242 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15243 altivec_vmulosw): New patterns.
15244 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15245 VMULOSW): Add definitions.
15247 2017-06-20 Julia Koval <julia.koval@intel.com>
15249 * config/i386/i386.c: Fix rounding expand for new pattern.
15250 * config/i386/subst.md: Fix pattern (parallel -> unspec).
15252 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
15254 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
15255 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
15257 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
15259 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
15262 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
15264 * config/aarch64/aarch64-cores.def: Rearrange to sort by
15265 architecture, then by implementer ID.
15266 * config/aarch64/aarch64-tune.md: Regenerate.
15268 2017-06-20 Richard Biener <rguenther@suse.de>
15270 PR middle-end/81097
15271 * fold-const.c (split_tree): Fold to type before negating.
15273 2017-06-20 David Malcolm <dmalcolm@redhat.com>
15275 * diagnostic-show-locus.c
15276 (selftest::test_fixit_deletion_affecting_newline): New function.
15277 (selftest::diagnostic_show_locus_c_tests): Call it.
15279 2017-06-20 Andreas Schwab <schwab@suse.de>
15282 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
15285 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15287 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
15288 __ARM_FEATURE_COPROC according to support.
15290 2017-06-20 Jakub Jelinek <jakub@redhat.com>
15292 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
15293 Rewritten to avoid overflow for > 32-bit pointers.
15296 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
15297 by removing enum keyword.
15298 (ubsan_type_descriptor): Likewise. Formatting fix.
15301 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
15302 splitter): Require TARGET_SSE2 in the condition.
15304 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
15307 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
15308 for doing vector set of SFmode on ISA 3.0.
15309 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
15310 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
15312 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
15313 SFmode value into a V4SF variable that was extracted from another
15314 V4SF variable without converting the element to double precision
15315 and back to single precision vector format.
15316 (vsx_insert_extract_v4sf_p9_2): Likewise.
15318 2017-06-19 Jakub Jelinek <jakub@redhat.com>
15320 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
15321 in UWHI to avoid undefined overflow.
15324 * ubsan.h (enum ubsan_encode_value_phase): New.
15325 (ubsan_encode_value): Change second argument to
15326 enum ubsan_encode_value_phase with default value of
15327 UBSAN_ENCODE_VALUE_GENERIC.
15328 * ubsan.c (ubsan_encode_value): Change second argument to
15329 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
15330 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
15331 create_tmp_var_raw instead of create_tmp_var and use a
15333 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
15334 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
15335 ubsan_encode_value callers.
15338 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
15339 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
15340 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
15342 2017-06-19 Richard Biener <rguenther@suse.de>
15344 PR middle-end/81118
15345 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
15346 estimates if we changed anything.
15348 2017-06-19 Richard Biener <rguenther@suse.de>
15350 PR tree-optimization/80887
15351 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
15352 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
15353 simplified lookups, then reset mprts_hook.
15354 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
15356 (try_to_simplify): Likewise.
15358 2017-06-19 Martin Liska <mliska@suse.cz>
15361 * gimplify.c (gimplify_switch_expr):
15362 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
15364 2017-06-19 Martin Liska <mliska@suse.cz>
15366 * doc/install.texi: Document that PGO runs in 4 stages.
15368 2017-06-19 Martin Liska <mliska@suse.cz>
15371 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
15372 to dispatcher function name.
15373 * multiple_target.c (replace_function_decl): New function.
15374 (create_dispatcher_calls): Redirect both edges and references.
15376 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
15378 * profile-count.c (profile_count::dump): Dump quality.
15379 (profile_count::differs_from_p): Update for unsigned val.
15380 * profile-count.h (profile_count_quality): New enum.
15381 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
15383 2017-06-19 Richard Biener <rguenther@suse.de>
15385 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
15386 struct function as arg.
15387 (estimate_numbers_of_iterations): Export overload with loop arg.
15388 (free_numbers_of_iterations_estimates_loop): Use an overload of
15389 free_numbers_of_iterations_estimates instead.
15390 * tree-cfg.c (remove_bb): Adjust.
15391 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
15392 * tree-parloops.c (gen_parallel_loop): Likewise.
15393 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
15395 (tree_unroll_loops_completely): Likewise.
15396 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
15397 Use an overload instead and export.
15398 (estimated_loop_iterations): Adjust.
15399 (max_loop_iterations): Likewise.
15400 (likely_max_loop_iterations): Likewise.
15401 (estimate_numbers_of_iterations): Take struct function as arg
15403 (loop_exits_before_overflow): Adjust.
15404 (free_numbers_of_iterations_estimates_loop): Use an overload.
15405 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
15406 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
15408 2017-06-19 Richard Biener <rguenther@suse.de>
15411 * ipa-prop.c (find_constructor_constant_at_offset): Handle
15412 RANGE_EXPR conservatively.
15414 2017-06-16 Carl Love <cel@us.ibm.com>
15416 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15417 definitions for vec_float, vec_float2, vec_floato,
15418 vec_floate built-ins.
15419 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
15420 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
15422 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
15423 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
15424 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
15425 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
15426 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
15427 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
15428 vec_floato): Add builtin defines.
15429 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
15430 Update the built-in documentation file for the new built-in
15433 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15435 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
15436 (mthumb): Mark as the negative of -marm.
15438 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15440 * doc/invoke.texi (ARM Options, -mcpu): Document supported
15442 (ARM Options, -mtune): Document that this accepts the same
15443 extension options as -mcpu.
15444 (ARM Options, -mfpu): Document addition of -mfpu=auto.
15446 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15448 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
15449 permitted extensions.
15451 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15453 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
15454 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
15455 (armv8-m.main): Add option +nodsp.
15456 * config/arm/arm-cpu-cdata.h: Regenerated.
15458 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15460 * config/arm/t-fuchsia: New file.
15461 * config.gcc (arm*-*-fuchsia*): Use it.
15463 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15465 * config/arm/t-symbian: Rewrite for new option infrastructure.
15467 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15469 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
15470 (MULTILIB_REQUIRED): Likewise.
15472 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15474 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
15475 (MULTILIB_RESUE): Likewise.
15476 (MULTILIB_MATCHES): Likewise.
15477 (MULTLIB_REQUIRED): Likewise.
15479 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15481 * config/arm/t-rtems: Rewrite for new option framework.
15483 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15485 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
15486 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
15487 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
15488 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
15489 * config/arm/t-multilib: ... here.
15490 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
15491 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
15492 armv7-a and armv8*-a when A-profile libraries have not been built.
15493 * config/arm/t-rmprofile: Rewrite.
15495 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15497 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
15498 with a backslash. Remove the backslash after substituting unescaped
15500 * doc/fragments.texi (MULTILIB_REUSE): Document it.
15502 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15504 * config.gcc: (arm*-*-*): When building a-profile libraries, force
15505 the driver to pass through the default setting of -mfloat-abi.
15506 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
15508 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
15509 (all_feat_combs): New rule.
15510 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
15512 * config/arm/t-aprofile: Rewrite.
15514 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15516 * config/arm/arm.h (FPUTYPE_AUTO): Define.
15517 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
15518 fpu is not specified by the user/command-line.
15519 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
15520 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
15521 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
15522 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
15523 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
15524 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
15526 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15528 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
15529 * config/arm/t-arm-elf: Rewritten.
15531 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15533 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
15534 have some floating-point instructions.
15535 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
15536 (TARGET_MAYBE_HARD_FLOAT): New macro.
15537 * config/arm/arm-builtins.c (arm_init_builtins): Use
15538 TARGET_MAYBE_HARD_FLOAT.
15539 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
15541 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15543 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
15544 (configargs.h): Include it.
15545 (arm_print_hint_for_fpu_option): New function.
15546 (arm_parse_fpu_option): New function.
15547 (candidate_extension): New class.
15548 (arm_canon_for_multilib): New function.
15549 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
15550 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
15551 (ARCH_CANONICAL_SPECS): New macro.
15552 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
15554 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15556 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
15557 are set after handling multilib fragments. Set target_cpu_default2
15560 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15562 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
15564 (arm*-*-*): Set target_cpu_default2 to a quoted string.
15565 * config/arm/parsecpu.awk (check_cpu): Validate any extension
15567 (check_arch): Likewise.
15568 * config/arm/arm.c (arm_configure_build_target): Handle
15569 TARGET_CPU_DEFAULT being a string constant. Scan any feature
15570 options in the default.
15572 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15574 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
15575 when an option is an alias of another.
15576 * config/arm/parsecpu.awk (optalias): New parser token.
15577 (gen_comm_data): Mark non-alias options as such. Emit entries
15578 for extension aliases.
15579 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
15580 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
15581 (armv6kz, armv6zk, armv6t2): Likewise.
15582 (armv7): Make vfpv3-d16 an alias.
15583 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
15585 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
15586 Sort in canonical order.
15587 (armv8-a): Sort in canonical order.
15588 (armv8.1-a, armv8.2-a): Likewise.
15589 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
15591 (cortex-a9): Sort in canonical order.
15592 * config/arm/arm.c (selftests.h): Include it.
15593 (arm_test_cpu_arch_data): New function.
15594 (arm_run_self_tests): New function.
15595 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
15596 (targetm): Move declaration to the end of the file.
15597 * arm-cpu-cdata.h: Regenerated.
15599 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15601 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
15602 call to target_mode_check describing the type of option passed.
15603 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
15604 (arm_target_thumb_only): Use arm_parse_arch_option_name or
15605 arm_parse_cpu_option_name to match parameters against list of
15607 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
15608 arm_arch_core_flags data structure.
15609 * config/arm/arm-cpu_cdata.h: Regenerated.
15611 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15613 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
15615 (arm_print_hint_for_cpu_option): Likewise.
15616 (arm_print_hint_for_arch_option): Likewise.
15617 (arm_parse_cpu_option_name): Likewise.
15618 (arm_parse_arch_option_name): Likewise.
15619 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
15620 of entries in the all_fpus list.
15621 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
15622 (arm_parse_cpu_option_name): Declare.
15623 (arm_parse_arch_option_name): Declare.
15624 (arm_parse_option_features): Declare.
15625 (arm_intialize_isa): Declare.
15626 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
15628 (gen_comm_data): ... here. Make definitions non-static.
15629 * config/arm/arm-cpu-data.h: Regenerated.
15630 * config/arm/arm-cpu-cdata.h: Regenerated.
15632 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15634 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
15635 (cpu_arch_extension): New structure.
15636 (cpu_arch_option, arch_option, cpu_option): New structures.
15637 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
15638 architecture types.
15639 (gen_data): Generate new format data tables.
15640 * config/arm/arm.c (cpu_tune): New structure.
15641 (cpu_option, processors): Delete.
15642 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
15643 (arm_print_hint_for_cpu_option): ... this and ...
15644 (arm_print_hint_for_arch_option): ... this.
15645 (arm_parse_arch_cpu_name): Delete. Replace with ...
15646 (arm_parse_cpu_option_name): ... this and ...
15647 (arm_parse_arch_option_name): ... this.
15648 (arm_unrecognized_feature): Change type of target parameter to
15650 (arm_parse_arch_cpu_features): Delete. Replace with ...
15651 (arm_parse_option_features): ... this.
15652 (arm_configure_build_target): Rework to use new configuration data
15654 (arm_print_tune_info): Rework for new configuration data tables.
15655 * config/arm/arm-cpu-data.h: Regenerated.
15656 * config/arm/arm-cpu.h: Regenerated.
15658 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15660 * Makefile.in (OBJS): Move sbitmap.o from here ...
15661 (OBJS-libcommon): ... to here.
15663 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15665 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
15666 (ISA_ALL_CRYPTO): New macro.
15667 (ISA_ALL_SIMD): New macro
15668 (ISA_ALL_FP): New macro.
15669 * config/arm/arm.c (fpu_bitlist): Update initializer.
15670 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
15672 (arm9e): Add fpu. Add option for nofp
15673 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
15674 (arm926ej-s, arm1026ej-s): Likewise.
15675 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
15676 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
15677 neon-fp16, neon-vfpv4, nofp and nosimd.
15678 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
15679 (cortex-a8): Add fpu. Add option for nofp.
15680 (cortex-a9): Add fpu. Add options for nosimd and nofp.
15681 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
15682 (cortex-r4f): Add fpu.
15683 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
15684 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
15686 (cortex-r8): Likewise.
15687 (cortex-m4): Add fpu. Add option for nofp.
15688 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
15689 (cortex-a17.cortex-a7): Likewise.
15690 (cortex-a32): Add fpu. Add options for crypto and nofp.
15691 (cortex-a35, cortex-a53): Likewise.
15692 (cortex-a57): Add fpu. Add option for crypto.
15693 (cortex-a72, cortex-a73): Likewise.
15694 (exynos-m1): Likewise.
15695 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
15696 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
15697 (cortex-m33): Add fpu. Add option for nofp.
15698 * config/arm/arm-cpu-cdata.h: Regenerated
15699 * config/arm/arm-cpu-data.h: Regenerated.
15701 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15703 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
15704 (armv5te, armv5tej): Likewise.
15705 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
15706 (armv7): Add options fp and vfpv3-d16.
15707 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
15708 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
15710 (armv7ve): Likewise.
15711 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
15712 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
15713 (armv8-a): Add nocrypto option.
15714 (armv8.1-a, armv8.2-a): Likewise.
15715 (armv8-m.main): add options fp, fp.dp and nofp.
15717 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15719 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
15721 (armv8-a+crc): Delete.
15722 (armv8.1-a): Add options simd, crypto and nofp.
15723 (armv8.2-a): Add options fp16, simd, crypto and nofp.
15724 (armv8.2-a+fp16): Delete.
15725 (armv8-m.main): Add option dsp.
15726 (armv8-m.main+dsp): Delete.
15727 (cortex-a8): Add fpu. Add nofp option.
15728 (cortex-a9): Add fpu. Add nofp and nosimd options.
15729 * config/arm/parsecpu.awk (gen_data): Generate option tables and
15730 link to main cpu and architecture data structures.
15731 (gen_comm_data): Only put isa attributes from the main architecture
15733 (option): New statement for architecture and CPU entries.
15734 * arm.c (struct cpu_option): New structure.
15735 (struct processors): Add entry for options.
15736 (arm_unrecognized_feature): New function.
15737 (arm_parse_arch_cpu_name): Ignore any characters after the first
15739 (arm_parse_arch_cpu_feature): New function.
15740 (arm_configure_build_target): Separate out any CPU and architecture
15741 features and parse separately. Don't error out if -mfpu=auto is
15742 used with only an architecture string.
15743 (arm_print_asm_arch_directives): New function.
15744 (arm_file_start): Call it.
15745 * config/arm/arm-cpu-cdata.h: Regenerated.
15746 * config/arm/arm-cpu-data.h: Likewise.
15747 * config/arm/arm-tables.opt: Likewise.
15749 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15751 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
15752 assembler when it is not -mfpu=auto.
15754 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15756 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
15757 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
15758 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
15759 (ASM_CPU_SPEC): Rewrite.
15760 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
15761 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
15762 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
15763 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
15764 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
15765 copied string is NUL-terminated. Also strip any characters prefixed
15767 (arm_rewrite_selected_arch): New function.
15768 (arm_rewrite_march): New function.
15770 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15772 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
15773 (x_arm_cpu_string, x_arm_tune_string): Likewise.
15774 (march, mcpu, mtune): Convert to string-based options.
15775 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
15776 (arm_parse_arch_cpu_name): New function.
15777 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
15778 identify selected architecture or CPU.
15779 (arm_option_save): New function.
15780 (TARGET_OPTION_SAVE): Redefine.
15781 (arm_option_restore): Restore string options.
15782 (arm_option_print): Print string options.
15784 2017-06-16 Martin Sebor <msebor@redhat.com>
15786 PR tree-optimization/80933
15787 PR tree-optimization/80934
15788 * builtins.c (fold_builtin_3): Do not handle bcmp here.
15789 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
15790 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
15791 (gimple_fold_builtin): Call them.
15793 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15795 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
15796 as unlikely; update profile.
15798 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15800 * predict.c (force_edge_cold): Handle declaring edges impossible
15803 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15805 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
15807 (try_unroll_loop_completely): Fix reporting.
15809 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15811 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
15813 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
15816 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
15817 if given a non-constant argument for an intrinsic which requires a
15820 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15822 * profile.c (compare_freqs): New function.
15823 (branch_prob): Sort edge list.
15824 (find_spanning_tree): Assume that the list is priority sorted.
15826 2017-06-16 Richard Biener <rguenther@suse.de>
15828 PR tree-optimization/81090
15829 * passes.def (pass_record_bounds): Remove.
15830 * tree-pass.h (make_pass_record_bounds): Likewise.
15831 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
15832 make_pass_record_bounds): Likewise.
15833 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
15834 not free niter estimates at the beginning but at the end.
15835 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
15837 2017-06-16 Richard Biener <rguenther@suse.de>
15839 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
15840 initializer to workaround ICE in host GCC 4.8.
15842 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15844 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
15846 (clone_inlined_nodes): Update.
15848 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15850 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
15851 prefetch settings, and enable prefetching by default at -O3.
15853 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15855 * config/aarch64/aarch64.c (aarch64_override_options_internal):
15856 Set flag_prefetch_loop_arrays according to tuning data.
15858 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15860 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
15861 New tune structure.
15862 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
15863 [Unrelated to main purpose of the patch] Place the pointer field last
15864 to enable type checking errors when tune structure are wrongly merged.
15865 * config/aarch64/aarch64.c (generic_prefetch_tune,)
15866 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
15867 (thunderx2t99_prefetch_tune): New tune constants.
15868 (tune_params *_tunings): Update all tunings (no functional change).
15869 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
15870 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
15871 from tunings structures.
15873 2017-06-16 Jakub Jelinek <jakub@redhat.com>
15876 * ubsan.c (instrument_null): Add T argument, use it instead
15877 of computing it based on IS_LHS.
15878 (instrument_object_size): Likewise.
15879 (pass_ubsan::execute): Adjust instrument_null and
15880 instrument_object_size callers to pass gimple_get_lhs or
15881 gimple_assign_rhs1 result to it. Use instrument_null instead of
15882 calling get_base_address and instrument_mem_ref. Handle
15883 aggregate call arguments for object-size sanitization.
15885 2017-06-16 Yury Gribov <tetra2005@gmail.com>
15887 PR tree-optimization/81089
15888 * tree-vrp.c (is_masked_range_test): Validate operands of
15891 2017-06-15 Martin Sebor <msebor@redhat.com>
15894 * dumpfile.c (dump_register): Avoid calling memset to initialize
15895 a class with a default ctor.
15896 * gcc.c (struct compiler): Remove const qualification.
15897 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
15898 * hash-table.h: Ditto.
15899 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
15901 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
15902 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
15904 * params.h (struct param_info): Make struct members non-const.
15905 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
15906 with default initialization.
15907 * vec.h (vec_copy_construct, vec_default_construct): New helper
15909 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
15910 with vec_copy_construct.
15911 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
15912 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
15913 * doc/invoke.texi (-Wclass-memaccess): Document.
15915 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15917 * emit-rtl.h (is_leaf): Update comment about local
15918 register allocator.
15920 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15923 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
15924 for a variable to have a section before checking if the section has a
15926 Set section to.persistent if persistent attribute is set.
15927 Warn if .persistent attribute is used on an automatic variable.
15929 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
15931 PR rtl-optimization/80474
15932 * reorg.c (update_block): Do not ignore instructions in a delay slot.
15934 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
15936 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
15939 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
15941 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
15942 (casesi): Emit bounds checking as RTL.
15943 (casesi_internal_mips16_<mode>): Remove bounds checking.
15945 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
15947 * config/xtensa/xtensa.c (xtensa_option_override): Append
15948 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
15949 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
15950 xtensa_doloop_hooks): Define unconditionally.
15951 (xtensa_reorg_loops): Only call reorg_loops in the presence of
15953 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
15954 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
15955 for it in xtensa_option_override.
15956 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
15957 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
15959 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
15961 * doc/cppopts.texi: Document '-' special value to -MF.
15963 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
15965 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
15966 (cortex_a53_fconst): Likewise.
15967 (cortex_a53_fpmul): Likewise.
15968 (cortex_a53_f_load_64): Likewise.
15969 (cortex_a53_f_load_many): Likewise.
15970 (cortex_a53_advsimd_alu): Likewise.
15971 (cortex_a53_advsimd_alu_q): Likewise.
15972 (cortex_a53_advsimd_mul): Likewise.
15973 (cortex_a53_advsimd_mul_q): Likewise.
15974 (fpmac bypass): Add new bypass for fpmac-fpmac case.
15975 Add missing fmul, r2f_cvt and fconst cases.
15977 2017-06-14 Richard Biener <rguenther@suse.de>
15979 PR middle-end/81088
15980 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
15982 (fold_binary_loc): When associating do not treat pre-existing
15983 TREE_OVERFLOW on literal constants as a reason to allow
15984 TREE_OVERFLOW on associated literal constants.
15986 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
15988 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
15989 (MASK_FEATURES): New macro.
15990 * config/sparc/sparc.c (sparc_option_override): Remove the special
15991 handling of -mfpu and generalize it to all MASK_FEATURES switches.
15993 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
15995 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
15996 a division of 0 if non-call exceptions are enabled.
15998 2017-06-14 Andrew Pinski <apinski@cavium.com>
15999 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16002 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
16003 Improve vector initialization code gen for only variable case.
16005 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
16007 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
16009 2017-06-14 Richard Biener <rguenther@suse.de>
16011 PR tree-optimization/81083
16012 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
16015 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16017 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
16018 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
16019 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
16020 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
16021 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
16022 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
16024 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16026 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
16027 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
16029 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16031 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
16033 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16035 * config/rs6000/t-rtems: Don't handle SPE.
16037 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16039 * config/rs6000/t-linux: Don't handle SPE.
16041 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16043 * config/rs6000/eabispe.h: Delete file.
16045 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16047 * config/rs6000/t-spe: Delete file.
16049 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16051 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
16052 (rs6000_legitimate_offset_address_p): Return false for anything in
16053 V2SImode or V2SFmode.
16055 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16057 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
16058 except V2SF and V2SI. Rearrange the vector modes, and add comments.
16059 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
16061 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
16062 (rs6000_legitimate_offset_address_p): Ditto.
16063 (rs6000_emit_move): Ditto.
16064 (rs6000_init_builtins): Remove V4HI_type_node.
16066 2017-06-13 Martin Liska <mliska@suse.cz>
16069 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
16070 (gate_asan): Likewise.
16071 * asan.h (asan_no_sanitize_address_p): Remove the function.
16072 (sanitize_flags_p): New function.
16073 * builtins.def: Fix coding style.
16074 * common.opt: Use renamed enum value.
16075 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
16076 * doc/extend.texi: Document no_sanitize attribute.
16077 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
16078 to SANITIZE_UNDEFINED_NONDEFAULT.
16079 * gcc.c (sanitize_spec_function): Use the renamed enum value.
16080 * gimple-fold.c (optimize_atomic_compare_exchange_p):
16081 Use sanitize_flags_p.
16082 * gimplify.c (gimplify_function_tree): Likewise.
16083 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
16084 * opts.c (parse_no_sanitize_attribute): New function.
16085 (common_handle_option): Use renamed enum value.
16086 * opts.h (parse_no_sanitize_attribute): Declare.
16087 * tree.c (sanitize_flags_p): New function.
16088 * tree.h: Declared here.
16089 * tsan.c: Use sanitize_flags_p.
16090 * ubsan.c (ubsan_expand_null_ifn): Likewise.
16091 (instrument_mem_ref): Likewise.
16092 (instrument_bool_enum_load): Likewise.
16093 (do_ubsan_in_current_function): Remove the function.
16094 (pass_ubsan::execute): Use sanitize_flags_p.
16095 * ubsan.h: Remove do_ubsan_in_current_function
16096 * tree-cfg.c (print_no_sanitize_attr_value): New function.
16097 (dump_function_to_file): Use it here.
16099 2017-06-13 Martin Jambor <mjambor@suse.cz>
16101 PR tree-optimization/80803
16102 PR tree-optimization/81063
16103 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
16104 (propagate_subaccesses_across_link): Enqueue subtree whenever
16105 necessary instead of relying on the caller.
16107 2017-06-13 Martin Jambor <mjambor@suse.cz>
16109 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
16110 that have a first_link.
16111 (sort_and_splice_var_accesses): Do not check first_link before
16113 (subtree_mark_written_and_enqueue): Likewise.
16114 (propagate_all_subaccesses): Likewise and do not stop at first
16115 parent with a first_link.
16117 2017-06-13 Martin Jambor <mjambor@suse.cz>
16119 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
16122 2017-06-13 Yury Gribov <tetra2005@gmail.com>
16124 * match.pd: New pattern.
16126 2017-06-13 Yury Gribov <tetra2005@gmail.com>
16128 * tree-vrp.c (is_masked_range_test): New function.
16129 (register_edge_assert_for): Determine ranges for
16132 2017-06-13 Yury Gribov <tetra2005@gmail.com>
16134 PR tree-optimization/67328
16135 * fold-const.c (maskable_range_p): New function.
16136 (build_range_check): Generate bittests if possible.
16138 2017-06-13 Martin Liska <mliska@suse.cz>
16140 * gimple-pretty-print.c (dump_probability): Add new argument.
16141 (dump_edge_probability): Dump both probability and count.
16142 (dump_gimple_label): Likewise.
16143 (dump_gimple_bb_header): Likewise.
16145 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
16148 * config/avr/avr-devices.c: Fix indentation.
16149 * config/avr/gen-avr-mmcu-specs.c: Dito.
16151 2017-06-13 Richard Biener <rguenther@suse.de>
16153 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
16154 instead get vector type from stmt_info.
16155 (vectorizable_reduction): Adjust. Remove dead code.
16157 2017-06-13 Richard Biener <rguenther@suse.de>
16159 PR middle-end/81065
16160 * fold-const.c (extract_muldiv_1): Remove bogus distribution
16161 case of C * (x * C2 + C3).
16162 (fold_addr_of_array_ref_difference): Properly fold index difference.
16164 2017-06-12 David S. Miller <davem@davemloft.net>
16167 * config/sparc/sparc.md (return expander): Emit frame blockage if
16168 function uses alloca.
16170 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
16172 * combine.c (make_field_assignment): Check len rather than the mode
16173 precision when calling force_to_mode.
16175 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
16177 Support multilibs and devices that see flash in RAM address range.
16180 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
16181 (avr_mcu_t) <flash_pm_offset>: New field.
16182 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
16183 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
16184 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
16185 (AVR_TINY_PM_OFFSET): Remove macro.
16186 * config/avr/avr.opt (-mshort-calls): New option.
16187 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
16188 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
16189 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
16190 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
16191 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
16192 instead of avr_arch->have_jmp_call.
16193 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
16194 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
16195 avr_arch->flash_pm_offset to define.
16196 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
16197 new field flash_pm_offset. Add entry for avrxmega3.
16198 (avr_texinfo): Add entry for avrxmega3.
16199 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
16200 attiny212, attiny214,
16201 attiny412, attiny414, attiny416, attiny417,
16202 attiny814, attiny816, attiny817,
16203 attiny1614, attiny1616, attiny1617,
16204 attiny3214, attiny3216, attiny3217.
16205 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
16206 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
16207 (avr_print_operand_address) [AVR_TINY]: Same.
16208 (avr_asm_init_sections) <readonly_data_section>: Only patch
16209 callback if avr_arch->flash_pm_offset = 0.
16210 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
16211 for rodata if avr_arch->flash_pm_offset != 0.
16212 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
16213 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
16214 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
16215 (m_options): Append opt_rcall.
16216 (m_dirnames): Append dir_rcall.
16217 * config/avr/t-multilib: Regenerate.
16219 * configure.ac [target=avr]: Check whether avrxmega3 default
16220 linker description file works as needed.
16221 * configure: Regenerate.
16222 * doc/avr-mmcu.texi: Regenerate.
16223 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
16224 <__AVR_ARCH__>: Document avrxmega3 and 103.
16225 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
16226 <__AVR_SHORT_CALLS__>: Document it.
16227 <__AVR_PM_BASE_ADDRESS__>: Document it.
16228 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
16229 (AVR Variable Attributes) <progmem>: Document this is
16230 not needed for avrxmega3.
16231 (AVR Named Address Spaces) <__flash>: Dito.
16233 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
16235 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
16237 2017-06-12 Doug Rupp <rupp@adacore.com>
16239 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
16240 Append vxworks-stdint.h to the tm_file list.
16241 * config/vxworks-stdint.h: New file.
16243 2017-06-12 Martin Liska <mliska@suse.cz>
16245 PR tree-optimization/81041
16246 * tree-profile.c (gimple_gen_ic_func_profiler):
16247 Create an extra BB in profile-generate
16248 (gimple_gen_time_profiler): Likewise.
16250 2017-06-12 Jakub Jelinek <jakub@redhat.com>
16252 PR tree-optimization/81003
16253 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
16254 (update_range_test): Use it instead of force_gimple_operand_gsi.
16256 2017-06-12 Richard Biener <rguenther@suse.de>
16258 PR tree-optimization/81053
16259 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
16260 with backedge value not defined in loop. Simplify def stmt
16263 2017-06-11 Tom de Vries <tom@codesourcery.com>
16266 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
16268 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
16269 nvptx_cannot_force_const_mem.
16271 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16273 * opts.c (finish_options): Move test for flag_split_stack after
16274 it has been initialized.
16276 2017-06-11 Jason Merrill <jason@redhat.com>
16278 * tree.h (id_equal): New.
16279 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
16280 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
16281 instead of strcmp of IDENTIFIER_POINTER.
16283 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16285 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
16286 (mark_all_inlined_calls_cdtor): Fix formating.
16287 (inline_transform): Rescale profile before inlining.
16289 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16291 * cgraph.h (cgraph_edge::clone): Update prototype.
16292 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
16293 (cgraph_node::create_clone): Update.
16294 (cgraph_node::create_version_clone): Update.
16295 * tree-inline.c (copy_bb): Update.
16296 (expand_call_inline): Update.
16298 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
16300 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
16301 factored out from ...
16302 (rs6000_emit_prologue): ... here.
16304 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
16306 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
16307 factored out from ...
16308 (rs6000_emit_prologue): ... here.
16310 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16312 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
16314 (handle_missing_profiles): Fix computation of tp_first_run.
16315 (counts_to_freqs): Do not touch freqs when count is 0.
16317 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16319 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
16322 2017-06-10 Tom de Vries <tom@codesourcery.com>
16324 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
16325 attributes): Document signal effective target.
16327 2017-06-10 Tom de Vries <tom@codesourcery.com>
16329 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
16330 Document effective target stack_size.
16332 2017-06-09 David Malcolm <dmalcolm@redhat.com>
16334 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
16335 to the edit_context if they can be auto-applied.
16337 2017-06-9 Ian Lance Taylor <iant@golang.org>
16339 * opts.c (finish_options): If -fsplit-stack, disable implicit
16340 -forder-blocks-and-partition.
16341 * doc/invoke.texi (Optimize Options): Document that when using
16342 -fsplit-stack -forder-blocks-and-partition is not implicitly
16345 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16347 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
16348 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
16349 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
16350 * builtins.def (abort, trap, unreachable): Declare cold.
16351 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
16352 * tree-core.h (ECF_COLD): New.
16353 * tree.c (set_call_expr_flags): Handle ECF_COLD.
16354 (build_common_builtin_nodes): Mark unreachable and abort as cold.
16356 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16358 * predict.c (unlikely_executed_stmt_p): Cleanup.
16360 2017-06-09 Richard Biener <rguenther@suse.de>
16362 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
16363 model if the ref is always written to.
16365 2017-06-09 Tamar Christina <tamar.christina@arm.com>
16367 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
16369 2017-06-09 Tamar Christina <tamar.christina@arm.com>
16371 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
16374 2017-06-09 Tom de Vries <tom@codesourcery.com>
16377 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
16378 "target cannot support label values" when encountering LABEL_REF.
16380 2017-06-09 Martin Liska <mliska@suse.cz>
16382 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
16383 (gimple_gen_ic_func_profiler): Emit direct comparison
16384 of __gcov_indirect_call_callee with NULL.
16385 (gimple_gen_time_profiler): Change probability from
16386 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
16388 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16390 * profile.c (edge_gcov_counts): Turn to pointer.
16391 (compute_branch_probabilities, compute_branch_probabilities): Update.
16392 (branch_prob): Do not clear edge_gcov_count.
16393 * profile.h (edge_gcov_counts): Turn to pointer.
16394 (edge_gcov_count): Update.
16396 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16398 * gimple.h (gimple_check_failed): Mark cold.
16400 2017-06-09 Richard Biener <rguenther@suse.de>
16402 PR tree-optimization/66623
16403 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
16404 refactor check_reduction into two parts, properly computing
16405 whether we have to check reduction validity for outer loop
16408 2017-06-09 Richard Biener <rguenther@suse.de>
16410 PR tree-optimization/79483
16411 * graphite-scop-detection.c (order): New global.
16412 (get_order): Compute bb to order mapping that satisfies code
16413 generation constraints.
16414 (cmp_pbbs): New helper.
16415 (build_scops): Start domwalk at entry block, sort generated
16418 2017-06-09 Richard Biener <rguenther@suse.de>
16420 PR middle-end/81007
16421 * ipa-polymorphic-call.c
16422 (ipa_polymorphic_call_context::restrict_to_inner_class):
16423 Skip FIELD_DECLs with error_mark_node type.
16424 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
16427 2017-06-09 Martin Liska <mliska@suse.cz>
16429 * predict.c (struct branch_predictor): New struct.
16430 (test_prediction_value_range): New test.
16431 (predict_c_tests): New function.
16432 * selftest-run-tests.c (selftest::run_tests): Run the function.
16433 * selftest.h: Declare new tests.
16435 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
16438 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
16439 gen_add3_insn did not fail.
16440 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
16441 r0, construct that number in a temporary reg and add that reg to r0.
16442 If asked to put the result in r0 as well, fail.
16444 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
16446 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
16447 for early expansion of vec_eqv.
16449 2017-06-08 Jakub Jelinek <jakub@redhat.com>
16451 PR middle-end/81005
16452 * ubsan.c (instrument_null): Avoid pointless code temporary.
16453 (pass_ubsan::execute): Instrument aggregate arguments of calls.
16455 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
16459 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
16462 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
16465 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16467 * predict.c (unlikely_executed_edge_p): Move ahead.
16468 (probably_never_executed_edge_p): Use it.
16470 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
16472 PR middle-end/79988
16473 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
16474 gimple_call_builtin_p call.
16476 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16478 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
16479 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
16480 rtl_check_failed_type2, rtl_check_failed_code1,
16481 rtl_check_failed_code2, rtl_check_failed_code_mode,
16482 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
16483 rtvec_check_failed_bounds, rtl_check_failed_flag,
16484 _fatal_insn_not_found, _fatal_insn): Likewise.
16485 * tree.h (tree_contains_struct_check_failed,
16486 tree_check_failed, tree_not_check_failed,
16487 tree_class_check_failed, tree_range_check_failed,
16488 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
16489 tree_vec_elt_check_failed, phi_node_elt_check_failed,
16490 tree_operand_check_failed, omp_clause_check_failed,
16491 omp_clause_operand_check_failed, omp_clause_range_check_failed):
16494 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16496 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
16497 flag_branch_probabilities.
16498 * ipa-inline.c (edge_badness): Likewise.
16499 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
16500 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
16501 * predict.c (maybe_hot_frequency_p): Likewise.
16502 (probably_never_executed): Likewise.
16503 * sched-ebb.c (schedule_ebbs): Likewise.
16504 * sched-rgn.c (find_single_block_region): Likewise.
16505 * tracer.c (tail_duplicate): Likewise.
16507 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16509 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
16510 longer requires x_flag_profile_use.
16512 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16514 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
16515 instead of flag_reorder_blocks_and_partition.
16516 * dbxout.c (dbxout_function_end): Likewise.
16517 * dwarf2out.c (gen_subprogram_die): Likewise.
16518 * haifa-sched.c (sched_create_recovery_edges): Likewise.
16519 * hw-doloop.c (reorg_loops): Likewise.
16520 * varasm.c (assemble_start_function,
16521 assemble_end_function): Likewise.
16522 (decide_function_section): Do not check for
16523 flag_reorder_blocks_and_partition.
16525 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
16527 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
16529 (chkp_get_hard_register_fake_addr_expr): Ditto.
16530 (chkp_build_addr_expr): Add check for hard reg case.
16531 (chkp_parse_array_and_component_ref): Ditto.
16532 (chkp_find_bounds_1): Ditto.
16533 (chkp_process_stmt): Don't generate bounds store for
16536 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16538 * predict.c (maybe_hot_bb_p): Do not check profile status.
16539 (maybe_hot_edge_p): Likewise.
16540 (probably_never_executed): Check for zero counts even if profile
16542 (unlikely_executed_edge_p): New function.
16543 (unlikely_executed_stmt_p): New function.
16544 (unlikely_executed_bb_p): New function.
16545 (set_even_probabilities): Use unlikely predicates.
16546 (combine_predictions_for_bb): Likewise.
16547 (predict_paths_for_bb): Likewise.
16548 (predict_paths_leading_to_edge): Likewise.
16549 (determine_unlikely_bbs): New function.
16550 (estimate_bb_frequencies): Use it.
16551 (compute_function_frequency): Use zero counts even if profile is
16553 * profile-count.h: Fix typo.
16555 2017-08-08 Julia Koval <julia.koval@intel.com>
16557 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
16558 _mm512_mask_cvtsepi16_storeu_epi8,
16559 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
16560 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
16561 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
16562 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
16563 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
16564 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
16565 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
16566 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
16567 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
16568 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
16569 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
16570 __builtin_ia32_pmovuswb256mem_mask,
16571 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
16572 __builtin_ia32_pmovwb512mem_mask): New builtins.
16574 2017-08-08 Julia Koval <julia.koval@intel.com>
16576 PR target/73350,80862
16577 * config/i386/subst.md (round): Fix round pattern.
16578 * config/i386/i386.c (ix86_erase_embedded_rounding):
16579 Fix erasing rounding for the fixed pattern.
16581 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16583 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
16585 2017-06-08 Martin Liska <mliska@suse.cz>
16587 PR gcov-profile/80911
16588 * gcov.c (block_info::block_info): New constructor.
16590 2017-06-07 Carl Love <cel@us.ibm.com>
16592 * config/rs6000/rs6000-c: The return type of the following
16593 built-in functions was implemented as int not long long. Fix sign
16594 of return value for the unsigned version of vec_mulo and vec_mule.
16595 vector unsigned long long vec_bperm (vector unsigned long long,
16596 vector unsigned char)
16597 vector signed long long vec_mule (vector signed int,
16599 vector unsigned long long vec_mule (vector unsigned int,
16600 vector unsigned int)
16601 vector signed long long vec_mulo (vector signed int,
16603 vector unsigned long long vec_mulo (vector unsigned int,
16604 vector unsigned int)
16605 * doc/extend.texi: Fix the documentation for the built-in
16608 2017-06-07 Carl Love <cel@us.ibm.com>
16611 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
16614 2017-06-07 Carl Love <cel@us.ibm.com>
16616 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
16617 support, Generate doublehv for signed int/float for BE case only.
16619 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
16621 * doc/invoke.texi (mcx16): Rewrite.
16623 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16625 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
16626 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
16627 *mov<mode>_softfloat, and an anonymous splitter): Use
16628 nonimmediate_operand instead of rs6000_nonimmediate_operand.
16630 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16632 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
16634 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
16635 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
16636 (rs6000_debug_reg_global): Adjust.
16637 (rs6000_init_hard_regno_mode_ok): Adjust.
16638 (rs6000_dbx_register_number): Adjust.
16639 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
16640 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
16641 Remove SPE_ACC and SPEFSCR.
16642 (REG_ALLOC_ORDER): Ditto.
16643 (FRAME_POINTER_REGNUM): Change to 111.
16644 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
16645 (REG_CLASS_NAMES): Ditto.
16646 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
16647 (REGISTER_NAMES): Ditto.
16648 (ADDITIONAL_REG_NAMES): Ditto.
16649 (rs6000_reg_names): Ditto.
16650 * config/rs6000/rs6000.md: Renumber some register number
16653 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16655 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
16657 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
16658 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
16660 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
16661 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
16662 Delete the SPE high registers.
16663 (REG_ALLOC_ORDER): Ditto.
16664 (enum reg_class): Remove SPE_HIGH_REGS.
16665 (REG_CLASS_NAMES): Ditto.
16666 (REG_CLASS_CONTENTS): Delete the SPE high registers.
16667 (REGISTER_NAMES): Ditto.
16668 (rs6000_reg_names): Ditto.
16669 * doc/tm.texi.in: Remove SPE as example.
16670 * doc/tm.texi: Regenerate.
16672 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16674 * config/rs6000/8540.md (ppc8540_brinc): Delete.
16675 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
16676 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
16677 * config/rs6000/rs6000.md (type): Remove "brinc".
16679 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16681 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
16682 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
16683 * config/rs6000/linuxspe.h: Delete file.
16684 * config/rs6000/rs6000.md: Don't include spe.md.
16685 * config/rs6000/spe.h: Delete file.
16686 * config/rs6000/spe.md: Delete file.
16687 * config/rs6000/t-rs6000: Remove spe.md.
16689 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16691 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
16692 (reg_or_none500mem_operand): Delete.
16693 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
16694 instead of reg_or_none500mem_operand.
16696 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16698 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
16699 handling of SPE flags.
16700 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
16702 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16704 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
16706 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
16707 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
16708 paired_divv2sf3): Similar.
16709 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
16710 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
16711 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
16713 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
16714 Rename the paired_* instruction patterns.
16715 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
16717 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
16718 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
16719 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
16720 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
16721 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
16722 PAIRED_VECTOR_MODE.
16723 (struct machine_function): Delete field spe_insn_chain_scanned_p.
16724 (spe_func_has_64bit_regs_p): Delete.
16725 (spe_expand_predicate_builtin): Delete.
16726 (spe_expand_evsel_builtin): Delete.
16727 (TARGET_DWARF_REGISTER_SPAN): Do not define.
16728 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
16729 (invalid_e500_subreg): Delete.
16730 (rs6000_legitimize_address): Always force_reg op2 as well, for
16731 paired single memory accesses.
16732 (rs6000_member_type_forces_blk): Delete.
16733 (rs6000_spe_function_arg): Delete.
16734 (rs6000_expand_unop_builtin): Delete SPE handling.
16735 (rs6000_expand_binop_builtin): Ditto.
16736 (spe_expand_stv_builtin): Delete.
16737 (bdesc_2arg_spe): Delete.
16738 (spe_expand_builtin): Delete.
16739 (spe_expand_predicate_builtin): Delete.
16740 (spe_expand_evsel_builtin): Delete.
16741 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
16742 (spe_init_builtins): Delete.
16743 (spe_func_has_64bit_regs_p): Delete.
16744 (savres_routine_name): Delete "info" parameter. Adjust callers.
16745 (rs6000_emit_stack_reset): Ditto.
16746 (rs6000_dwarf_register_span): Delete.
16747 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
16748 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
16749 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
16751 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
16753 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
16754 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
16755 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
16756 mulv2sf3, divv2sf3): Delete expanders.
16758 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16760 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
16762 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16764 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
16765 * config/rs6000/rs6000.c: Ditto.
16767 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16769 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
16770 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
16771 comparison_operator.
16773 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16775 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
16776 * config/rs6000/rs6000.opt: Ditto.
16777 * config/rs6000/t-rtems: Ditto.
16779 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16781 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
16782 TARGET_E500_SINGLE by 0, simplify.
16783 * config/rs6000/rs6000.c: Ditto.
16784 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
16785 (spe_build_register_parallel): Delete.
16786 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
16787 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
16788 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
16789 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
16790 (E500_CONVERT): Delete.
16791 * config/rs6000/spe.md: Remove many patterns and all define_constants.
16793 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16795 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
16796 * config/rs6000/dfp.md: Ditto.
16797 (negdd2, *negdd2_fpr): Merge.
16798 (absdd2, *absdd2_fpr): Merge.
16799 (negtd2, *negtd2_fpr): Merge.
16800 (abstd2, *abstd2_fpr): Merge.
16801 * config/rs6000/e500.h: Delete file.
16802 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
16803 TARGET_FPRS by 1 and simplify.
16804 * config/rs6000/rs6000-c.c: Ditto.
16805 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
16806 TARGET_DF_SPE by 0.
16807 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
16809 * config/rs6000/rs6000.md: Ditto.
16810 (floatdidf2, *floatdidf2_fpr): Merge.
16811 (move_from_CR_gt_bit): Delete.
16812 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
16813 (E500_CR_IOR_COMPARE): Delete.
16814 (All patterns that require !TARGET_FPRS): Delete.
16815 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
16817 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16819 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
16821 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16823 * graphds.c (add_edge): Intitialize edge's attached data.
16824 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
16825 pointer parameter. Call pointed function on each edge during
16826 graph traversing. Skip traversing the edge when the function
16828 (graphds_dfs, graphds_scc): Ditto.
16829 (for_each_edge): New parameter. Pass the new parameter to callback
16831 * graphds.h (skip_edge_callback): New function pointer type.
16832 (graphds_dfs, graphds_scc): New function pointer parameter.
16833 (graphds_edge_callback, for_each_edge): New parameter.
16835 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16837 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
16838 out code checking if runtime alias check is possible to below ...
16839 Call the new function.
16840 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
16841 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
16843 2017-06-07 Marek Polacek <polacek@redhat.com>
16846 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
16847 TYPE_OVERFLOW_WRAPS check.
16849 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16851 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
16852 if versioning is required.
16853 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
16854 peeling with the check for versioning.
16856 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16858 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
16859 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
16860 Set true to new parameter if new ssa variable is defined.
16861 (vect_gen_vector_loop_niters): Refactor. Set range information
16862 for the new vector loop bound variable.
16863 (vect_do_peeling): Ditto.
16865 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16867 * tree-affine.c (ssa.h): Include header file.
16868 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
16869 has wrapping overflow behavior.
16871 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16873 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
16875 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16877 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
16878 (tree_to_aff_combination): ... here.
16880 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16882 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
16883 reg_pressure model function.
16884 (ivopts_global_cost_for_size): Delete.
16885 (determine_set_costs, iv_ca_recount_cost): Call new model function
16886 ivopts_estimate_reg_pressure.
16888 2017-06-07 Tamar Christina <tamar.christina@arm.com>
16890 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
16891 expensive than udiv. Remove floating point cases from mod.
16893 2017-06-07 Tamar Christina <tamar.christina@arm.com>
16895 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
16896 Increase idiv cost.
16898 2017-06-07 Tamar Christina <tamar.christina@arm.com>
16900 * config/aarch64/aarch64.md
16901 (copysignsf3): Fix mask generation.
16903 2017-06-07 Jakub Jelinek <jakub@redhat.com>
16905 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
16907 (class dump_manager): Add register_dumps method.
16908 * dumpfile.c: Include langhooks.h.
16909 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
16910 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
16911 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
16912 (dump_manager::dump_register): Start with 512 entries instead of 32.
16913 (dump_manager::register_dumps): New method.
16914 * toplev.c (general_init): Instead of invoking register_dumps
16915 langhook, invoke register_dumps method on the dump manager.
16916 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
16919 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
16921 * doc/md.texi: Clarify the restrictions on a define_insn condition.
16922 Say that # requires an associated define_split to exist, and that
16923 the define_split must be suitable for use after register allocation.
16925 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16927 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
16928 (compute_outgoing_frequencies): Also initialize zero counts.
16929 (find_many_sub_basic_blocks): Do not produce uninitialized profile
16930 around loops; preserve more of profile when nothing changes.
16932 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
16934 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
16936 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
16937 * config/arm/arm-cpu-cdata.h: Regenerate.
16938 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
16939 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
16940 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
16941 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
16942 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
16944 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
16946 * config/arm/t-rmprofile: Likewise.
16947 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
16949 2017-06-06 David S. Miller <davem@davemloft.net>
16952 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
16953 blockage if function uses alloca.
16955 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16957 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
16958 New "uid" fields to hold pretty-print IDs of group and ref.
16959 Memory references are now identified as <group_id>:<ref_id>
16960 instead of using [random] addresses.
16961 (dump_mem_details): Simplify, no functional change.
16962 (dump_mem_ref): Simplify and make output more concise.
16963 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
16964 (find_or_create_group): Initialize group uid.
16965 (record_ref): Initialize ref uid. Improve debug output.
16966 (prune_group_by_reuse, should_issue_prefetch_p,)
16967 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
16968 (mark_nontemporal_store, determine_loop_nest_reuse):
16969 Improve debug output.
16971 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16973 * dbgcnt.def (prefetch): New debug counter.
16974 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
16975 (schedule_prefetches): Stop issueing prefetches if debug counter
16978 2017-06-06 Tom de Vries <tom@codesourcery.com>
16980 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
16981 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
16983 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16985 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
16986 Use aarch64_reg_or_zero predicate for operand 4.
16987 (aarch64_compare_and_swap<mode> define_insn_and_split):
16988 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
16989 (aarch64_store_exclusive<mode>): Likewise for operand 2.
16991 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
16993 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
16994 (arm_compute_save_core_reg_mask): This.
16995 (thumb1_compute_save_reg_mask): Rename into ...
16996 (thumb1_compute_save_core_reg_mask): This.
16997 (arm_compute_save_reg0_reg12_mask): Adapt comment.
16998 (arm_compute_frame_layout): Likewise.
17000 2017-06-06 Richard Biener <rguenther@suse.de>
17002 PR tree-optimization/80974
17003 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
17004 keep or clear leaders SSA info.
17006 2017-06-06 Tom de Vries <tom@codesourcery.com>
17008 * config/nvptx/nvptx.c (split_mode_p): New function.
17009 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
17011 2017-06-06 Tom de Vries <tom@codesourcery.com>
17013 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
17015 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17018 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
17021 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17023 * shrink-wrap.c (handle_simple_exit): Update profile.
17024 (try_shrink_wrapping): Upate profile.
17026 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17028 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
17029 (tree_guess_outgoing_edge_probabilities): New.
17030 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
17031 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
17033 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17035 * ipa-split.c (split_function): Initialize return bb profile.
17037 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17039 * profile.c (compute_branch_probabilities): Also initialize
17040 EXIT_BLOCK profile.
17042 2017-06-06 Richard Biener <rguenther@suse.de>
17044 PR tree-optimization/80928
17045 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
17046 (vect_analyze_loop_operations): Properly guard analysis for
17048 (vect_transform_loop): Likewise.
17049 (vect_analyze_loop_2): Also reset SLP type on PHIs.
17050 (vect_model_induction_cost): Do not cost for pure SLP.
17051 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
17052 of induction in inner loop vectorization.
17053 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
17054 (vect_get_and_check_slp_defs): Handle vect_induction_def.
17055 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
17057 (vect_analyze_slp_cost_1): Cost induction.
17058 (vect_detect_hybrid_slp_stmts): Handle PHIs.
17059 (vect_get_slp_vect_defs): Likewise.
17060 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
17061 (vect_transform_stmt): Handle SLP reductions.
17062 * tree-vectorizer.h (vectorizable_induction): Adjust.
17064 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
17066 * config/rs6000/rs6000.c (make_resolver_func): Update
17067 init_lowered_empty_function call.
17069 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
17071 * doc/invoke.texi: Document the -fprofile-abs-path option.
17072 * common.opt (fprofile-abs-path): New option.
17073 * gcov-io.h (gcov_write_filename): Declare.
17074 * gcov-io.c (gcov_write_filename): New function.
17075 * coverage.c (coverage_begin_function): Use gcov_write_filename.
17076 * profile.c (output_location): Likewise.
17078 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17080 * shring-wrap.c: Revert accidental commit.
17082 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
17084 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
17086 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17088 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
17090 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
17091 profile in callgraph edge.
17092 * profile-count.h (apply_probability): If THIS is 0, then result is 0
17093 (apply_scale): Likewise.
17094 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
17095 Also scale profile when inlining function with zero profile.
17096 (initialize_cfun): Update exit block profile even when it is zero.
17097 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
17098 when profile is read.
17100 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
17102 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
17103 (CLONE_*): New constants to define the processors we can generate
17104 code for with the target_clone attribute.
17105 (rs6000_clone_map): New array to identify which clone processors
17106 the current program is running on.
17107 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
17108 target_clone attribute.
17109 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
17110 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
17111 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
17112 (cpu_expand_builtin): Add support for target_clone attribute.
17113 (rs6000_valid_attribute_p): Allow "default" attribute.
17114 (get_decl_name): New debug function to simplify printing the
17115 current function name in debugging statements.
17116 (rs6000_clone_priority): New functions to support the target_clone
17117 attribute, and be able to generate code to switch between ISA 2.05
17118 through ISA 3.0 (power6 through power9).
17119 (rs6000_compare_version_priority): Likewise.
17120 (rs6000_get_function_versions_dispatcher): Likewise.
17121 (make_resolver_func): Likewise.
17122 (add_condition_to_bb): Likewise.
17123 (dispatch_function_versions): Likewise.
17124 (rs6000_generate_version_dispatcher_body): Likewise.
17125 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
17126 (fusion_gpr_load_p): Fix a spacing issue.
17127 * doc/extend.texi (Common Function Attributes): Document that the
17128 PowerPC supports the target_clone attribute.
17130 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
17132 * config/arm/arm.h: explain F symbol found in description of ARM
17133 register allocation in its legend.
17135 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17137 * config/mips/frame-header-opt.c: Include profile-count.h.
17138 * config/riscv/riscv.c: Include profile-count.h
17140 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17142 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
17144 (sm_set_flag_if_changed): Add bbs field.
17145 (execute_sm_if_changed_flag_set): Pass BBS.
17146 (execute_sm): Update.
17148 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17150 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
17153 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17155 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
17156 (peephole2): New peephole2 to emit the above.
17157 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
17159 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17161 * config/aarch64/aarch64.c (define_peephole2 above
17162 *sub_<shift>_<mode>): New peephole.
17164 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17166 * config/i386/i386.c (make_resolver_func): Update.
17167 * Makefile.in: Add profile-count.h and profile-count.o
17168 * auto-profile.c (afdo_indirect_call): Update to new API.
17169 (afdo_set_bb_count): Update.
17170 (afdo_propagate_edge): Update.
17171 (afdo_propagate_circuit): Update.
17172 (afdo_calculate_branch_prob): Update.
17173 (afdo_annotate_cfg): Update.
17174 * basic-block.h: Include profile-count.h
17175 (struct edge_def): Turn count to profile_count.
17176 (struct basic_block_def): Likewie.
17177 (REG_BR_PROB_BASE): Move to profile-count.h
17178 (RDIV): Move to profile-count.h
17179 * bb-reorder.c (max_entry_count): Turn to profile_count.
17180 (find_traces): Update.
17181 (rotate_loop):Update.
17182 (connect_traces):Update.
17183 (sanitize_hot_paths):Update.
17184 * bt-load.c (migrate_btr_defs): Update.
17185 * cfg.c (RDIV): Remove.
17186 (init_flow): Use alloc_block.
17187 (alloc_block): Uninitialize count.
17188 (unchecked_make_edge): Uninitialize count.
17189 (check_bb_profile): Update.
17190 (dump_edge_info): Update.
17191 (dump_bb_info): Update.
17192 (update_bb_profile_for_threading): Update.
17193 (scale_bbs_frequencies_int): Update.
17194 (scale_bbs_frequencies_gcov_type): Update.
17195 (scale_bbs_frequencies_profile_count): New.
17196 * cfg.h (update_bb_profile_for_threading): Update.
17197 (scale_bbs_frequencies_profile_count): Declare.
17198 * cfgbuild.c (compute_outgoing_frequencies): Update.
17199 (find_many_sub_basic_blocks): Update.
17200 * cfgcleanup.c (try_forward_edges): Update.
17201 (try_crossjump_to_edge): Update.
17202 * cfgexpand.c (expand_gimple_tailcall): Update.
17203 (construct_exit_block): Update.
17204 * cfghooks.c (verify_flow_info): Update.
17205 (dump_bb_for_graph): Update.
17206 (split_edge): Update.
17207 (make_forwarder_block): Update.
17208 (duplicate_block): Update.
17209 (account_profile_record): Update.
17210 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
17211 (get_estimated_loop_iterations): Update.
17212 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
17213 (single_likely_exit): Update.
17214 * cfgloopmanip.c (scale_loop_profile): Update.
17216 (set_zero_probability): Update.
17217 (lv_adjust_loop_entry_edge): Update.
17218 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
17219 (purge_dead_edges): Update.
17220 (rtl_account_profile_record): Update.
17221 * cgraph.c (cgraph_node::create): Uninitialize count.
17222 (symbol_table::create_edge): Uninitialize count.
17223 (cgraph_update_edges_for_call_stmt_node): Update.
17224 (cgraph_edge::dump_edge_flags): Update.
17225 (cgraph_node::dump): Update.
17226 (cgraph_edge::maybe_hot_p): Update.
17227 * cgraph.h: Include profile-count.h
17228 (create_clone), create_edge, create_indirect_edge): Update.
17229 (cgraph_node): Turn count to profile_count.
17230 (cgraph_edge0: Likewise.
17231 (make_speculative, clone): Update.
17232 (create_edge): Update.
17233 (init_lowered_empty_function): Update.
17234 * cgraphclones.c (cgraph_edge::clone): Update.
17235 (duplicate_thunk_for_node): Update.
17236 (cgraph_node::create_clone): Update.
17237 * cgraphunit.c (cgraph_node::analyze): Update.
17238 (cgraph_node::expand_thunk): Update.
17239 * final.c (dump_basic_block_info): Update.
17240 * gimple-streamer-in.c (input_bb): Update.
17241 * gimple-streamer-out.c (output_bb): Update.
17242 * graphite.c (print_global_statistics): Update.
17243 (print_graphite_scop_statistics): Update.
17244 * hsa-brig.c: Include basic-block.h.
17245 * hsa-dump.c: Include basic-block.h.
17246 * hsa-gen.c (T sum_slice): Update.
17247 (convert_switch_statements):Update.
17248 * hsa-regalloc.c: Include basic-block.h.
17249 * ipa-chkp.c (chkp_produce_thunks): Update.
17250 * ipa-cp.c (struct caller_statistics): Update.
17251 (init_caller_stats): Update.
17252 (gather_caller_stats): Update.
17253 (ipcp_cloning_candidate_p): Update.
17254 (good_cloning_opportunity_p): Update.
17255 (get_info_about_necessary_edges): Update.
17256 (dump_profile_updates): Update.
17257 (update_profiling_info): Update.
17258 (update_specialized_profile): Update.
17259 (perhaps_add_new_callers): Update.
17260 (decide_about_value): Update.
17261 (ipa_cp_c_finalize): Update.
17262 * ipa-devirt.c (struct odr_type_warn_count): Update.
17263 (struct decl_warn_count): Update.
17264 (struct final_warning_record): Update.
17265 (possible_polymorphic_call_targets): Update.
17266 (ipa_devirt): Update.
17267 * ipa-fnsummary.c (redirect_to_unreachable): Update.
17268 * ipa-icf.c (sem_function::merge): Update.
17269 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
17270 * ipa-inline.c (compute_uninlined_call_time): Update.
17271 (compute_inlined_call_time): Update.
17272 (want_inline_small_function_p): Update.
17273 (want_inline_self_recursive_call_p): Update.
17274 (edge_badness): Update.
17275 (lookup_recursive_calls): Update.
17276 (recursive_inlining): Update.
17277 (inline_small_functions): Update.
17278 (dump_overall_stats): Update.
17279 (dump_inline_stats): Update.
17280 * ipa-profile.c (ipa_profile_generate_summary): Update.
17281 (ipa_propagate_frequency): Update.
17282 (ipa_profile): Update.
17283 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
17284 * ipa-utils.c (ipa_merge_profiles): Update.
17285 * loop-doloop.c (doloop_modify): Update.
17286 * loop-unroll.c (report_unroll): Update.
17287 (unroll_loop_runtime_iterations): Update.
17288 * lto-cgraph.c (lto_output_edge): Update.
17289 (lto_output_node): Update.
17290 (input_node): Update.
17291 (input_edge): Update.
17292 (merge_profile_summaries): Update.
17293 * lto-streamer-in.c (input_cfg): Update.
17294 * lto-streamer-out.c (output_cfg): Update.
17295 * mcf.c (create_fixup_graph): Update.
17296 (adjust_cfg_counts): Update.
17297 (sum_edge_counts): Update.
17298 * modulo-sched.c (sms_schedule): Update.
17299 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
17300 * predict.c (maybe_hot_count_p): Update.
17301 (probably_never_executed): Update.
17302 (dump_prediction): Update.
17303 (combine_predictions_for_bb): Update.
17304 (propagate_freq): Update.
17305 (handle_missing_profiles): Update.
17306 (counts_to_freqs): Update.
17307 (rebuild_frequencies): Update.
17308 (force_edge_cold): Update.
17309 * predict.h: Include profile-count.h
17310 (maybe_hot_count_p, counts_to_freqs): UPdate.
17311 * print-rtl-function.c: Do not include cfg.h
17312 * print-rtl.c: Include basic-block.h
17313 * profile-count.c: New file.
17314 * profile-count.h: New file.
17315 * profile.c (is_edge_inconsistent): Update.
17316 (correct_negative_edge_counts): Update.
17317 (is_inconsistent): Update.
17318 (set_bb_counts): Update.
17319 (read_profile_edge_counts): Update.
17320 (compute_frequency_overlap): Update.
17321 (compute_branch_probabilities): Update; Initialize and deinitialize
17323 (branch_prob): Update.
17324 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
17325 (edge_gcov_count): New.
17326 (bb_gcov_count): New.
17327 * shrink-wrap.c (try_shrink_wrapping): Update.
17328 * tracer.c (better_p): Update.
17329 * trans-mem.c (expand_transaction): Update.
17330 (ipa_tm_insert_irr_call): Update.
17331 (ipa_tm_insert_gettmclone_call): Update.
17332 * tree-call-cdce.c: Update.
17333 * tree-cfg.c (gimple_duplicate_sese_region): Update.
17334 (gimple_duplicate_sese_tail): Update.
17335 (gimple_account_profile_record): Update.
17336 (execute_fixup_cfg): Update.
17337 * tree-inline.c (copy_bb): Update.
17338 (copy_edges_for_bb): Update.
17339 (initialize_cfun): Update.
17340 (freqs_to_counts): Update.
17341 (copy_cfg_body): Update.
17342 (expand_call_inline): Update.
17343 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
17344 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
17345 (try_unroll_loop_completely): Update.
17346 (try_peel_loop): Update.
17347 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17348 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
17349 * tree-ssa-loop-split.c (connect_loops): Update.
17350 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
17351 * tree-ssa-reassoc.c (branch_fixup): Update.
17352 * tree-ssa-tail-merge.c (replace_block_by): Update.
17353 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
17354 (compute_path_counts): Update.
17355 (update_profile): Update.
17356 (recompute_probabilities): Update.
17357 (update_joiner_offpath_counts): Update.
17358 (estimated_freqs_path): Update.
17359 (freqs_to_counts_path): Update.
17360 (clear_counts_path): Update.
17361 (ssa_fix_duplicate_block_edges): Update.
17362 (duplicate_thread_path): Update.
17363 * tree-switch-conversion.c (case_bit_test_cmp): Update.
17364 (struct switch_conv_info): Update.
17365 * tree-tailcall.c (decrease_profile): Update.
17366 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
17367 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17368 * value-prof.c (check_counter): Update.
17369 (gimple_divmod_fixed_value): Update.
17370 (gimple_mod_pow2): Update.
17371 (gimple_mod_subtract): Update.
17372 (gimple_ic_transform): Update.
17373 (gimple_stringop_fixed_value): Update.
17374 * value-prof.h (gimple_ic): Update.
17376 2017-06-02 Carl Love <cel@us.ibm.com>
17378 * config/rs6000/rs6000-c: Add support for built-in functions
17379 vector double vec_doublee (vector signed int);
17380 vector double vec_doublee (vector unsigned int);
17381 vector double vec_doublee (vector float);
17382 vector double vec_doubleh (vector signed int);
17383 vector double vec_doubleh (vector unsigned int);
17384 vector double vec_doubleh (vector float);
17385 vector double vec_doublel (vector signed int);
17386 vector double vec_doublel (vector unsigned int);
17387 vector double vec_doublel (vector float);
17388 vector double vec_doubleo (vector signed int);
17389 vector double vec_doubleo (vector unsigned int);
17390 vector double vec_doubleo (vector float);.
17391 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
17392 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
17394 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
17395 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
17396 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
17398 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
17399 vec_doublel, vec_doubleh.
17400 * doc/extend.texi: Update the built-in documentation file for the
17401 new built-in functions.
17403 2017-06-02 David Malcolm <dmalcolm@redhat.com>
17406 * ipa-inline-analysis.c (free_growth_caches): Set
17407 edge_removal_hook_holder to NULL after removing it.
17409 2017-06-02 Sudakshina Das <sudi.das@arm.com>
17411 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
17412 comparision with zero.
17414 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
17415 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
17416 for early expansion of vec_min and vec_max builtins.
17417 (builtin_function_type): Add min/max unsigned variants to those
17418 identified as having unsigned arguments.
17420 2017-06-02 Olivier Hainque <hainque@adacore.com>
17422 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
17424 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17426 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
17427 Use VALL_F16 iterator rather than VALL.
17429 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17431 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
17432 Emit CBNZ inside loop when doing a strong exchange and comparing
17433 against zero. Generate the CC flags after the loop.
17435 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
17437 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
17438 (dl_section_ref): New.
17439 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
17440 On AIX, append an expression to subtract the size of the
17441 section length to dl_section_ref.
17443 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
17445 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
17446 for early expansion of vector absolute builtins.
17448 2017-06-02 Richard Biener <rguenther@suse.de>
17450 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
17451 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
17453 2017-06-02 Richard Biener <rguenther@suse.de>
17455 PR tree-optimization/80948
17456 * tree-tailcall.c (find_tail_calls): Track stmts to move in
17457 stmt order as well.
17459 2017-06-02 Richard Biener <rguenther@suse.de>
17461 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
17463 * tree-vect-stmts.c (process_use): Do not mark backedge defs
17464 for inductions as relevant.
17466 2017-06-02 Richard Biener <rguenther@suse.de>
17468 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
17469 (vectorizable_induction): ... this. Remove dead code.
17471 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
17473 * builtins. (expand_builtin_alloca): Remove second parameter and
17474 infer its value from the first parameter instead.
17475 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
17477 2017-06-02 Jakub Jelinek <jakub@redhat.com>
17479 PR rtl-optimization/80903
17480 * loop-doloop.c (add_test): Unshare sequence.
17482 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
17484 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
17486 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
17488 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
17490 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
17491 xlogue_layout::get_instance, logue_layout::xlogue_layout,
17492 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
17493 (xlogue_layout::get_stub_rtx): Make static.
17494 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
17495 (xlogue_layout::compute_stub_managed_regs): Rename to...
17496 (xlogue_layout::count_stub_managed_regs): ...this.
17497 (xlogue_layout::is_stub_managed_reg): New function.
17498 (xlogue_layout::m_stub_names): Rename to...
17499 (xlogue_layout::s_stub_names): ...this, make static.
17500 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
17501 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
17502 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
17503 xlogue_layout::s_stub_names): Instantiate statics.
17504 (stub_managed_regs): Remove.
17505 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
17506 (disable_call_ms2sysv_xlogues): Rename to...
17507 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
17508 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
17510 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
17511 change after reload_completed.
17512 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
17514 (ix86_expand_prologue): Likewise.
17515 (ix86_expand_epilogue): Likewise.
17516 (ix86_expand_split_stack_prologue): Likewise.
17517 (ix86_compute_frame_layout): Remove frame parameter ...
17518 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
17519 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
17521 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
17522 (ix86_frame): Move from here ...
17523 * config/i386/i386.h (ix86_frame): ... to here.
17524 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
17525 complete ix86_frame data structure instead. Remove some_ld_name.
17527 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
17529 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
17530 symbols that hold a DECL_VALUE_EXPR.
17532 2017-06-01 Martin Jambor <mjambor@suse.cz>
17534 PR tree-optimization/80898
17535 * tree-sra.c (process_subtree_disqualification): Removed.
17536 (disqualify_candidate): Do not acll
17537 process_subtree_disqualification.
17538 (subtree_mark_written_and_enqueue): New function.
17539 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
17540 RHS has been disqualified and re-queue LHS if necessary. Apart
17541 from that, ignore disqualified RHS.
17543 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17545 * config/s390/s390.c (s390_emit_epilogue): Disable early return
17546 address fetch for z10 or later.
17548 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17550 * config/arc/arc.md (tst_movb): Add guard when splitting.
17552 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17554 * config/arc/arc.c (arc_can_eliminate): Test against
17555 arc_frame_pointer_needed.
17557 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17559 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
17560 to prevent store reordering.
17561 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
17562 (type): Add block type.
17563 (stack_tie): Define special instruction to be used in
17566 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17568 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
17569 constraint. It is not valid for the pattern.
17570 (noncommutative_binary_comparison): Likewise.
17572 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17574 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
17577 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17579 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
17580 be used by the reg-alloc.
17582 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17584 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
17585 reg-alloc when having mul64 or mul32x16 instructions.
17586 (mulsidi3): Likewise.
17587 (umulsidi3): Likewise.
17588 (mulsi32x16): New pattern.
17589 (mulsi64): Likewise.
17590 (mulsidi64): Likewise.
17591 (umulsidi64): Likewise.
17592 (MUL32x16_REG): Define.
17593 (mul64_600): Use MUL32x16_REG.
17594 (mac64_600): Likewise.
17595 (umul64_600): Likewise.
17596 (umac64_600): Likewise.
17598 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17600 * config/arc/arc.md (mulsi3_700): Make it commutative.
17602 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
17604 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
17606 (*sign_extendsidi2_insn): Likewise for movstosw.
17608 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
17610 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
17611 the type of the input discriminant value. Convert the
17612 discriminant value of signedness vary.
17614 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
17616 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
17617 Add to -Wall section.
17619 2017-06-01 Richard Biener <rguenther@suse.de>
17621 PR middle-end/66313
17622 * fold-const.c (fold_plusminus_mult_expr): If the factored
17623 factor may be zero use a wrapping type for the inner operation.
17624 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
17625 and handle moved defs.
17626 (process_assignment): Properly guard the unary op case. Return a
17627 tri-state indicating that moving the stmt before the call may allow
17628 to continue. Pass through to_move.
17629 (find_tail_calls): Handle moving unrelated defs before
17632 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
17635 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
17636 splitter result in the canonical way.
17638 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
17640 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
17641 also for 32bit target. Update insn attributes.
17642 (zero-extendsidi2 splitter): Allow all registers for operand 1.
17644 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
17646 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
17647 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
17648 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
17649 (_mm_maskz_min_ss): New intrinsics.
17651 2017-05-31 Martin Liska <mliska@suse.cz>
17653 * tree-vect-loop.c (vect_create_epilog_for_reduction):
17654 Change comment style to one we normally use.
17655 (vectorizable_reduction): Likewise.
17656 (vectorizable_induction): Likewise.
17657 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
17658 (vectorizable_call): Likewise.
17659 (vectorizable_simd_clone_call): Likewise.
17660 (vectorizable_conversion): Likewise.
17661 (vectorizable_assignment): Likewise.
17662 (vectorizable_shift): Likewise.
17663 (vectorizable_operation): Likewise.
17664 (vectorizable_store): Likewise.
17665 (vectorizable_load): Likewise.
17666 * tree-vectorizer.h: Likewise.
17668 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
17670 * passes.c (emergency_dump_function): New.
17671 * tree-pass.h (emergency_dump_function): Declare.
17672 * plugin.c (plugins_internal_error_function): Remove.
17673 * plugin.h (plugins_internal_error_function): Remove declaration.
17674 * toplev.c (internal_error_function): New static function. Use it...
17675 (general_init): ...here.
17677 2017-05-31 Graham Markall <graham.markall@embecosm.com>
17679 * config/arc/arc.c (arc_print_operand): Handle constant operands.
17680 (arc_rtx_costs): Add costs for new patterns.
17681 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
17682 * config/arc/predicates.md: Add _1_2_3_operand predicate.
17684 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
17686 * tree-ssa-strlen.c (get_next_strinfo): New function.
17687 (get_stridx_plus_constant): Use it.
17688 (zero_length_string): Likewise.
17689 (adjust_related_strinfos): Likewise.
17690 (adjust_last_stmt): Likewise.
17692 2017-05-31 Richard Biener <rguenther@suse.de>
17695 * config/i386/i386.c (ix86_expand_builtin): Remove assert
17696 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
17698 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
17700 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
17701 loop_vinfo argument and use of dependence distance vectors.
17702 Check instead whether the two references differ only in their
17703 initial value and assume that they have the same alignment if the
17704 difference is a multiple of the vector alignment.
17705 (vect_analyze_data_refs_alignment): Update call accordingly.
17707 2017-05-31 Martin Liska <mliska@suse.cz>
17710 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
17712 2017-05-31 Bin Cheng <bin.cheng@arm.com>
17714 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
17715 (create_intersect_range_checks): Move from ...
17716 * tree-data-ref.c (create_intersect_range_checks_index)
17717 (create_intersect_range_checks): ... to here.
17718 (create_runtime_alias_checks): New function factored from ...
17719 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
17720 here. Call above function.
17721 * tree-data-ref.h (create_runtime_alias_checks): New function.
17723 2017-05-31 Bin Cheng <bin.cheng@arm.com>
17725 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
17726 segment length for dr_b and compute it in wide_int.
17728 2017-05-31 Richard Biener <rguenther@suse.de>
17730 PR tree-optimization/80906
17731 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
17732 and pass through iv_map.
17733 (copy_bb_and_scalar_dependences): Adjust.
17734 (translate_pending_phi_nodes): Likewise.
17735 (copy_loop_close_phi_args): Handle code-generating IVs instead
17738 2017-05-30 David Malcolm <dmalcolm@redhat.com>
17740 * diagnostic-color.c (color_dict): Add "type-diff".
17741 (parse_gcc_colors): Update comment.
17742 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
17743 -fdiagnostics-show-template-tree and -fno-elide-type.
17744 (GCC_COLORS): Add type-diff to example.
17746 (-fdiagnostics-show-template-tree): New.
17747 (-fno-elide-type): New.
17748 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
17749 the pp_format_decoder callback. Call any m_format_postprocessor's
17751 (pretty_printer::pretty_printer): Initialize
17752 m_format_postprocessor.
17753 (pretty_printer::~pretty_printer): Delete any
17754 m_format_postprocessor.
17755 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
17756 (class format_postprocessor): New class.
17757 (struct pretty_printer::format_decoder): Document the new parameters.
17758 (struct pretty_printer::m_format_postprocessor): New field.
17759 * tree-diagnostic.c (default_tree_printer): Update for new
17760 bool and const char ** params.
17761 * tree-diagnostic.h (default_tree_printer): Likewise.
17763 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
17765 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
17766 (lwa_operand): Delete rs6000_gen_cell_microcode test.
17767 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
17768 rs6000_gen_cell_microcode code.
17769 (rs6000_final_prescan_insn): Delete.
17770 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
17771 "warn-cell-microcode" entries.
17772 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
17773 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
17774 throughout. Change cc_reg_not_micro_cr0_operand to
17775 cc_reg_not_cr0_operand throughout.
17776 (*extendhi<mode>2_noload): Delete.
17777 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
17778 (mwarn-cell-microcode): Delete.
17779 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
17780 -mgen-cell-microcode and -mwarn-cell-microcode.
17782 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
17785 * config/i386/constraints.md (Yd): New constraint.
17787 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
17788 and (?Yd, r) alternatives. Update insn attributes.
17789 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
17790 and (?*Yd, r) alternatives. Update insn attributes.
17791 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
17793 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
17795 * gimplify.c (gimplify_modify_expr): Don't create a
17796 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
17799 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
17801 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
17803 2017-05-30 Richard Biener <rguenther@suse.de>
17805 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
17806 and reduc_def fields.
17807 (STMT_VINFO_REDUC_TYPE): New define.
17808 (STMT_VINFO_REDUC_DEF): Likewise.
17809 (vect_force_simple_reduction): Adjust prototype.
17810 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
17811 (vect_is_simple_reduction): Remove check_reduction argument.
17812 (vect_force_simple_reduction): Adjust and set
17813 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
17814 (vectorizable_reduction): Do not re-do reduction analysis
17815 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
17816 * tree-parloops.c (gather_scalar_reductions): Adjust.
17818 2017-05-30 Richard Biener <rguenther@suse.de>
17820 PR middle-end/80901
17821 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
17824 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
17826 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
17827 Introduce unknown_misalignment parameter and remove vf.
17828 (vect_peeling_hash_get_lowest_cost):
17829 Pass unknown_misalignment parameter.
17830 (vect_enhance_data_refs_alignment):
17831 Fix unsupportable data ref treatment.
17833 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17835 * tree-vect-data-refs.c (vect_get_data_access_cost):
17836 Workaround for SLP handling.
17837 (vect_enhance_data_refs_alignment):
17838 Compute costs for doing no peeling at all, compare to the best
17839 peeling costs so far and avoid peeling if cheaper.
17841 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17843 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
17844 Return peeling info and set costs to zero for unlimited cost
17846 (vect_enhance_data_refs_alignment): Also inspect all datarefs
17847 with unknown misalignment. Compute and costs for unknown
17848 misalignment, compare them to the costs for known misalignment
17849 and choose the cheapest for peeling.
17851 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17853 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
17854 (vect_get_peeling_costs_all_drs): Create function.
17855 (vect_peeling_hash_get_lowest_cost):
17856 Use vect_get_peeling_costs_all_drs.
17857 (vect_peeling_supportable): Create function.
17858 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
17860 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17862 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
17863 DR_HAS_NEGATIVE_STEP.
17864 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
17865 (vect_enhance_data_refs_alignment): Use.
17866 (vect_duplicate_ssa_name_ptr_info): Use.
17867 * tree-vectorizer.h (dr_misalignment): Use.
17868 (known_alignment_for_access_p): Use.
17870 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17873 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
17875 (has_section_name): New function.
17877 2017-05-30 Martin Liska <mliska@suse.cz>
17880 * auto-profile.c (get_function_decl_from_block): Fix
17883 2017-05-30 Richard Biener <rguenther@suse.de>
17885 PR middle-end/80876
17886 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
17888 2017-05-30 Martin Liska <mliska@suse.cz>
17890 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
17891 * dumpfile.h (struct dump_file_info): Remove ctors.
17893 2017-05-30 Martin Liska <mliska@suse.cz>
17895 * predict.def: Fix GNU coding style.
17897 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
17899 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
17900 Mark 'to' argument with ATTRIBUTE_UNUSED.
17902 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
17904 * config/xtensa/xtensa.c (xtensa_emit_call): Use
17905 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
17906 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
17909 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
17911 * doc/install.texi (Options specification): Restore entry of
17912 --enable-sjlj-exceptions.
17914 2017-05-27 Michael Eager <eager@eagercon.com>
17917 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
17919 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
17921 * config/microblaze/microblaze.h
17922 (FIXED_REGISTERS): Update in macro.
17923 (CALL_USED_REGISTERS): Update in macro.
17925 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
17927 * doc/install.texi: Add links to macOS binary distributions.
17929 2017-05-27 Jakub Jelinek <jakub@redhat.com>
17933 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17935 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
17937 2017-05-26 Martin Liska <mliska@suse.cz>
17939 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
17941 2017-05-26 Martin Liska <mliska@suse.cz>
17943 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
17944 always leading ';; '.
17945 (dump_bb_info): Likewise.
17946 (brief_dump_cfg): Likewise.
17947 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
17948 * dumpfile.c: Remove usage of TDF_VERBOSE.
17949 * dumpfile.h (enum dump_kind): Likewise.
17950 (dump_gimple_bb_header): Do not use TDF_COMMENT.
17951 * print-tree.c (debug_verbose): Remove.
17952 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
17953 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
17954 * tree-diagnostic.c (default_tree_printer): Replace
17955 TDF_DIAGNOSTIC with TDF_SLIM.
17957 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17959 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
17960 in parameter loop, rather than loop_vinfo.
17961 (create_intersect_range_checks): Ditto.
17962 (vect_create_cond_for_alias_checks): Update call to above functions.
17964 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17966 PR tree-optimization/80815
17967 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
17968 for merging runtime alias checks. Handle negative DR_STEPs.
17970 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17972 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
17974 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
17975 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
17976 out code pruning runtime alias checks.
17977 * tree-data-ref.c (prune_runtime_alias_test_list): New function
17978 factored out from above.
17979 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
17981 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
17983 (prune_runtime_alias_test_list): New decalaration.
17985 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17987 * tree-vect-data-refs.c (compare_tree): Rename and move ...
17988 * tree-data-ref.c (data_ref_compare_tree): ... to here.
17989 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
17990 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
17991 (operator==, comp_dr_with_seg_len_pair): Ditto.
17992 (vect_prune_runtime_alias_test_list): Ditto.
17994 2017-05-26 Martin Liska <mliska@suse.cz>
17997 * params.def: Bound partial-inlining-entry-probability param.
17999 2017-05-26 Marek Polacek <polacek@redhat.com>
18002 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
18005 2017-05-26 Richard Biener <rguenther@suse.de>
18007 PR tree-optimization/80842
18008 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
18011 2017-05-26 Richard Biener <rguenther@suse.de>
18013 PR tree-optimization/80844
18014 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
18016 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
18018 * doc/md.texi (Machine Constraints): Update x86 family
18019 machine constraints section to match 'config/i386/constraints.md'.
18021 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
18023 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
18025 2017-05-25 Nathan Sidwell <nathan@acm.org>
18027 * doc/invoke.texi (--enable-languages): Update documentation.
18029 2017-05-25 Martin Liska <mliska@suse.cz>
18031 * dumpfile.c: Add TDF_FOLDING.
18032 * dumpfile.h (enum dump_kind): Likewise.
18033 * genmatch.c (dt_simplify::gen_1): Use it.
18035 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18037 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
18039 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18041 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
18042 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
18044 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18046 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
18047 * match.pd (X == C): Rewrite it here.
18048 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
18049 with_certain_nonzero_bits2): New predicates.
18050 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
18052 2017-05-24 Nathan Sidwell <nathan@acm.org>
18054 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
18057 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
18060 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
18062 * config/powerpcspe: New port. Files are copied from the rs6000
18063 port, with "rs6000" in filenames replaced by "powerpcspe".
18065 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
18067 PR rtl-optimization/80754
18068 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
18070 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
18072 * config/sparc/sparc.md (length): Return the correct value for -mflat
18073 sibcalls to match output_sibcall.
18075 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
18079 * config/rs6000/rs6000.c (struct machine_function): Add new field
18081 (rs6000_get_separate_components): Init that field, use it.
18082 (rs6000_components_for_bb): Use the field.
18084 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
18086 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
18088 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
18090 PR middle-end/80823
18091 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
18093 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18096 * config/s390/s390.c (s390_check_qrst_address): Check incoming
18097 address against address_operand predicate.
18098 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
18100 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
18102 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
18103 parameters passed indirectly.
18105 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
18107 * config/i386/i386.md (*movdi_internal): Remove SSE4
18108 alternative 18 (?r, *v). Update insn attributes.
18109 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
18110 Update insn attributes.
18111 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
18112 Update insn attributes.
18113 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
18114 alternative 1 (r, v). Remove isa attribute.
18115 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
18116 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
18117 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
18119 2017-05-23 Tom de Vries <tom@codesourcery.com>
18121 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
18124 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18126 * cgraphunit.c (symbol_table::process_new_functions): Update.
18127 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
18128 (inline_generate_summary): Rename to ...
18129 (ipa_fn_summary_generate): ... this one.
18130 (inline_read_summary): Rename to ...
18131 (ipa_fn_summary_read): ... this one.
18132 (inline_write_summary): Rename to ...
18133 (ipa_fn_summary_write): ... this one.
18134 (inline_free_summary): Rename to ...
18135 (ipa_free_fn_summary): ... this one.
18136 (pass_data_local_fn_summary, pass_local_fn_summary,
18137 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
18138 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
18139 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
18140 make_pass_ipa_fn_summary): New.
18141 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
18142 inline_write_summary, inline_free_summary): Remove.
18143 (ipa_free_fn_summary) : New.
18144 * ipa-inline.c (ipa_inline): Update.
18145 (pass_ipa_inline): Do not generate summaries.
18146 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
18148 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
18149 and add pass_ipa_fn_summary.
18150 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
18152 (make_pass_inline_parameters): Remove.
18154 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
18156 * omp-low.c (struct omp_context): Remove "default_kind" member.
18159 * omp-offload.c (execute_oacc_device_lower): Remove the
18160 parallelism dimensions function attributes for unparallelized
18161 OpenACC kernels constructs.
18163 2017-05-23 Martin Liska <mliska@suse.cz>
18165 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
18167 (cgraph_edge::make_speculative): Likewise.
18168 (cgraph_edge::resolve_speculation): Likewise.
18169 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
18170 (cgraph_node::dump): Likewise.
18171 * cgraph.h: Likewise.
18172 * cgraphunit.c (analyze_functions): Likewise.
18173 (symbol_table::compile): Likewise.
18174 * ipa-cp.c (print_all_lattices): Likewise.
18175 (determine_versionability): Likewise.
18176 (initialize_node_lattices): Likewise.
18177 (ipcp_verify_propagated_values): Likewise.
18178 (estimate_local_effects): Likewise.
18179 (update_profiling_info): Likewise.
18180 (create_specialized_node): Likewise.
18181 (perhaps_add_new_callers): Likewise.
18182 (decide_about_value): Likewise.
18183 (decide_whether_version_node): Likewise.
18184 (identify_dead_nodes): Likewise.
18185 (ipcp_store_bits_results): Likewise.
18186 * ipa-devirt.c (dump_targets): Likewise.
18187 (ipa_devirt): Likewise.
18188 * ipa-icf.c (sem_item::dump): Likewise.
18189 (sem_function::equals): Likewise.
18190 (sem_variable::equals): Likewise.
18191 (sem_item_optimizer::read_section): Likewise.
18192 (sem_item_optimizer::execute): Likewise.
18193 (congruence_class::dump): Likewise.
18194 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
18195 (dump_inline_summary): Likewise.
18196 (estimate_node_size_and_time): Likewise.
18197 (inline_analyze_function): Likewise.
18198 * ipa-inline-transform.c (inline_call): Likewise.
18199 * ipa-inline.c (report_inline_failed_reason): Likewise.
18200 (want_early_inline_function_p): Likewise.
18201 (edge_badness): Likewise.
18202 (update_edge_key): Likewise.
18203 (inline_small_functions): Likewise.
18204 * ipa-profile.c (ipa_profile): Likewise.
18205 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
18206 (ipa_make_edge_direct_to_target): Likewise.
18207 (remove_described_reference): Likewise.
18208 (ipa_impossible_devirt_target): Likewise.
18209 (propagate_controlled_uses): Likewise.
18210 (ipa_print_node_params): Likewise.
18211 (ipcp_transform_function): Likewise.
18212 * ipa-pure-const.c (pure_const_read_summary): Likewise.
18213 (propagate_pure_const): Likewise.
18214 * ipa-reference.c (generate_summary): Likewise.
18215 (read_write_all_from_decl): Likewise.
18216 (propagate): Likewise.
18217 (ipa_reference_read_optimization_summary): Likewise.
18218 * ipa-utils.c (ipa_merge_profiles): Likewise.
18219 * ipa.c (walk_polymorphic_call_targets): Likewise.
18220 (symbol_table::remove_unreachable_nodes): Likewise.
18221 (ipa_single_use): Likewise.
18222 * passes.c (execute_todo): Likewise.
18223 * predict.c (drop_profile): Likewise.
18224 * symtab.c (symtab_node::get_dump_name): New function.
18225 (symtab_node::dump_name): Likewise.
18226 (symtab_node::dump_asm_name): Likewise.
18227 (symtab_node::dump_references): Likewise.
18228 (symtab_node::dump_referring): Likewise.
18229 (symtab_node::dump_base): Likewise.
18230 (symtab_node::debug_symtab): Likewise.
18231 * tree-sra.c (convert_callers_for_node): Likewise.
18232 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
18233 * value-prof.c (init_node_map): Likewise.
18235 2017-05-23 Martin Liska <mliska@suse.cz>
18237 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
18238 and symtab_node::debug_symtab to symbol_table::debug.
18239 * cgraphunit.c (analyze_functions): Use the renamed function.
18240 (symbol_table::compile): Likewise.
18241 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
18242 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
18243 * passes.c (execute_todo): Likewise.
18244 * symtab.c (symbol_table::dump): New function.
18245 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
18247 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18249 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
18250 that nonconst implies exec.
18252 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18254 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
18255 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
18256 (inline_edge_summary_vec): Turn into ...
18257 (ipa_call_summaries): ... this one.
18258 (redirect_to_unreachable, edge_set_predicate,
18259 evaluate_properties_for_edge, inline_summary_alloc,
18260 reset_ipa_call_summary, reset_inline_summary,
18261 inline_summary_t::duplicate): Update.
18262 (inline_edge_duplication_hook): Turn to ...
18263 (ipa_call_summary_t::duplicate): ... this one.
18264 (inline_edge_removal_hook): Turn to ...
18265 (ipa_call_summary_t::remove): ... this one.
18266 (dump_inline_edge_summary): Turn to ...
18267 (dump_ipa_call_summary): ... this one.
18268 (estimate_function_body_sizes): Update.
18269 (inline_update_callee_summaries): Update.
18270 (remap_edge_change_prob): Update.
18271 (remap_edge_summaries): Update.
18272 (inline_merge_summary): Update.
18273 (do_estimate_edge_time): Update.
18274 (inline_generate_summary): Update.
18275 (inline_read_section): Update.
18276 (inline_read_summary): Update.
18277 (inline_free_summary): Update.
18278 * ipa-inline.c (can_inline_edge_p): Update.
18279 (compute_inlined_call_time): Update.
18280 (want_inline_small_function_p): Update.
18281 (edge_badness): Update.
18282 (early_inliner): Update.
18283 * ipa-inline.h (inline_edge_summary): Turn to ...
18284 (ipa_call_summary): ... this one.
18285 (ipa_call_summary_t): New class.
18286 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
18287 (ipa_call_summaries): New.
18288 (inline_edge_summary): Remove.
18289 (estimate_edge_growth): Update.
18290 * ipa-profile.c (ipa_propagate_frequency_1): Update.
18291 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
18292 * ipa-split.c (execute_split_functions): Update.
18293 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
18295 2017-05-23 Tom de Vries <tom@codesourcery.com>
18297 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
18298 attributes): Document rdrand effective target.
18300 2017-05-23 Tom de Vries <tom@codesourcery.com>
18302 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
18303 attributes): Sort alphabetically.
18305 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
18307 * config/avr/genmultilib.awk: Use gsub instead of gensub.
18309 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
18312 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
18313 V2DF/V2DI splat into two separate patterns, one that handles
18314 registers, and the other that only handles memory. Drop support
18315 for splatting from a GPR on ISA 2.07 and then splitting the
18316 splat into direct move and splat.
18317 (vsx_splat_<mode>_reg): Likewise.
18318 (vsx_splat_<mode>_mem): Likewise.
18320 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
18322 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
18324 2017-05-22 Jakub Jelinek <jakub@redhat.com>
18326 PR middle-end/80809
18327 * omp-low.c (finish_taskreg_remap): New function.
18328 (finish_taskreg_scan): If unit size of ctx->record_type
18329 is non-constant, unshare the size expression and replace
18330 decls in it with possible outer var refs.
18332 PR middle-end/80809
18333 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
18334 GOVD_SHARED rather than GOVD_PRIVATE with it.
18335 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
18336 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
18338 PR middle-end/80853
18339 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
18340 as last argument to build_outer_var_ref for pointer bases of array
18341 section reductions.
18343 2017-05-19 Martin Sebor <msebor@redhat.com>
18345 * print-tree.c (print_node): Print DECL_READ_P flag.
18347 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18349 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
18350 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
18351 * cgraph.c: Likewise.
18352 * cgraphunit.c: Likewise.
18353 * gengtype.c: Likewise.
18354 * ipa-cp.c: Likewise.
18355 * ipa-devirt.c: Likewise.
18356 * ipa-icf.c: Likewise.
18357 * ipa-predicate.c: Likewise.
18358 * ipa-profile.c: Likewise.
18359 * ipa-prop.c: Likewise.
18360 * ipa-split.c: Likewise.
18362 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
18363 edge_predicate_pool, dump_inline_hints,
18364 inline_summary::account_size_time, redirect_to_unreachable,
18365 edge_set_predicate, set_hint_predicate,
18366 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
18367 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
18368 inline_summary_t::remove, remap_hint_predicate_after_duplication,
18369 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
18370 ipa_call_summary_t::remove, initialize_growth_caches,
18371 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
18372 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
18373 mark_modified, unmodified_parm_1, unmodified_parm,
18374 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
18375 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
18376 compute_bb_predicates, will_be_nonconstant_expr_predicate,
18377 will_be_nonconstant_predicate, record_modified_bb_info,
18378 get_minimal_bb, record_modified, param_change_prob,
18379 phi_result_unknown_predicate, predicate_for_phi_result,
18380 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
18381 estimate_function_body_sizes, compute_inline_parameters,
18382 compute_inline_parameters_for_curren, pass_data_inline_parameters,
18383 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
18384 inline_update_callee_summaries, remap_edge_change_prob,
18385 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
18386 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
18387 inline_analyze_function, inline_summary_t::insert,
18388 inline_generate_summary, read_ipa_call_summary, inline_read_section,
18389 inline_read_summary, write_ipa_call_summary, inline_write_summary,
18390 inline_free_summary): Move to ipa-fnsummary.h
18391 (predicate_t): Remove.
18392 * ipa-fnsummary.c: New file.
18393 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
18394 (enum inline_hints_vals, inline_hints, agg_position_info,
18395 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
18396 inline_summaries, ipa_call_summary, ipa_call_summary_t,
18397 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
18398 dump_inline_summary, dump_inline_hints, inline_generate_summary,
18399 inline_read_summary, inline_write_summary, inline_free_summary,
18400 inline_analyze_function, initialize_inline_failed,
18401 inline_merge_summary, inline_update_overall_summary,
18402 compute_inline_parameters): Move to ipa-fnsummary.h
18403 * ipa-fnsummary.h: New file.
18404 * ipa-inline-transform.h: Include ipa-inline.h.
18405 * ipa-inline.c: LIkewise.
18407 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18409 * ipa-inline.c (edge_badness): Use inlined_time instead of
18410 inline_summaries->get.
18412 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18414 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
18416 2017-05-22 Nathan Sidwell <nathan@acm.org>
18418 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
18419 (fdump-lang): Document 'raw' option.
18420 * dumpfile.h (TDI_tu): Delete.
18421 * dumpfile.c (dump_files): Remove translation-unit.
18422 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
18424 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
18426 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
18427 command option from $(AWK) call.
18428 * config/avr/genmultilib.awk: Simplify and rewrite so that it
18429 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
18430 [FORMAT]: Remove handling of variable.
18431 * config/avr/t-multilib: Regenerate.
18433 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18435 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
18437 (dump_inline_summary): Do not print self_time.
18438 (estimate_function_body_sizes): Do not set self_time.
18439 (compute_inline_parameters): Likewise.
18440 (inline_read_section, inline_write_summary): Do not stream self_time.
18441 * ipa-inline.h (inline_summary): Drop self_time.
18443 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18445 * ipa-inline-analysis.c (account_size_time): Rename to ...
18446 (inline_summary::account_size_time): ... this one.
18447 (reset_ipa_call_summary): Turn to ...
18448 (ipa_call_summary::reset): ... this one.
18449 (reset_inline_summary): Turn to ...
18450 (inline_summary::reset): ... this one.
18451 (inline_summary_t::remove): Update.
18452 (inline_summary_t::duplicate): Update.
18453 (ipa_call_summary_t::remove): Update.
18454 (dump_inline_summary): Update.
18455 (estimate_function_body_sizes): Update.
18456 (compute_inline_parameters): Update.
18457 (estimate_node_size_and_time): Update.
18458 (inline_merge_summary): Update.
18459 (inline_update_overall_summary): Update.
18460 (inline_read_section): Update.
18461 (inline_write_summary): Update.
18462 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
18463 add account_size_time and reset member functions.
18464 (ipa_call_summary): Add reset function.
18465 * ipa-predicate.h (predicate::operator &): Constify.
18467 2017-05-22 Richard Biener <rguenther@suse.de>
18469 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
18471 2017-05-19 Jason Merrill <jason@redhat.com>
18473 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
18475 2017-05-19 Marek Polacek <polacek@redhat.com>
18478 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
18479 TYPE_OVERFLOW_WRAPS checks.
18481 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
18483 * tree-core.h (enum omp_clause_default_kind): Add
18484 "OMP_CLAUSE_DEFAULT_PRESENT".
18485 * tree-pretty-print.c (dump_omp_clause): Handle it.
18486 * gimplify.c (enum gimplify_omp_var_data): Add
18487 "GOVD_MAP_FORCE_PRESENT".
18488 (gimplify_adjust_omp_clauses_1): Map it to
18489 "GOMP_MAP_FORCE_PRESENT".
18490 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
18492 * gimplify.c (oacc_default_clause): Clarify.
18494 2017-05-19 Nathan Sidwell <nathan@acm.org>
18496 LANG_HOOK_REGISTER_DUMPS
18497 * toplev.c (general_init): Call register dump lang hook.
18498 * doc/invoke.texi: Document -fdump-lang option family.
18499 * dumpfile.c (dump_files): Remove class dump here.
18500 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
18501 * dumpfile.h (tree_dump_index): Remove TDI_class.
18502 * langhooks-def.h (lhd_register_dumps): Declare.
18503 (LANG_HOOKS_REGISTER_DUMPS): Define.
18504 (LANG_HOOKS_INITIALIZER): Add it.
18505 * langhooks.c (lhd_register_dumps): Define.
18506 * langhooks.h (struct lang_hooks): Add register_dumps.
18508 2017-05-19 Nathan Sidwell <nathan@acm.org>
18510 * context.h (context::set_passes): New.
18511 * context.c (context::context): Do not create pass manager.
18512 * toplev.c (general_init): Create pass manager here.
18514 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
18516 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
18517 use this splitter if two add or or instructions would also work for
18518 the constant we want to generate.
18520 2017-05-19 Richard Biener <rguenther@suse.de>
18523 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
18524 predicate evaluation.
18526 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
18528 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
18530 * ipa-inline.c (want_inline_small_function_p): Do not cast to
18533 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
18535 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
18536 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
18537 (inline_edge_summary_vec): Turn into ...
18538 (ipa_call_summaries): ... this one.
18539 (redirect_to_unreachable, edge_set_predicate,
18540 evaluate_properties_for_edge, inline_summary_alloc,
18541 reset_ipa_call_summary, reset_inline_summary,
18542 inline_summary_t::duplicate): Update.
18543 (inline_edge_duplication_hook): Turn to ...
18544 (ipa_call_summary_t::duplicate): ... this one.
18545 (inline_edge_removal_hook): Turn to ...
18546 (ipa_call_summary_t::remove): ... this one.
18547 (dump_inline_edge_summary): Turn to ...
18548 (dump_ipa_call_summary): ... this one.
18549 (estimate_function_body_sizes): Update.
18550 (inline_update_callee_summaries): Update.
18551 (remap_edge_change_prob): Update.
18552 (remap_edge_summaries): Update.
18553 (inline_merge_summary): Update.
18554 (do_estimate_edge_time): Update.
18555 (inline_generate_summary): Update.
18556 (inline_read_section): Update.
18557 (inline_read_summary): Update.
18558 (inline_free_summary): Update.
18559 * ipa-inline.c (can_inline_edge_p): Update.
18560 (compute_inlined_call_time): Update.
18561 (want_inline_small_function_p): Update.
18562 (edge_badness): Update.
18563 (early_inliner): Update.
18564 * ipa-inline.h (inline_edge_summary): Turn to ...
18565 (ipa_call_summary): ... this one.
18566 (ipa_call_summary_t): New class.
18567 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
18568 (ipa_call_summaries): New.
18569 (inline_edge_summary): Remove.
18570 (estimate_edge_growth): Update.
18571 * ipa-profile.c (ipa_propagate_frequency_1): Update.
18572 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
18573 * ipa-split.c (execute_split_functions): Update.
18574 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
18576 2017-05-19 Richard Biener <rguenther@suse.de>
18578 PR middle-end/80764
18579 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
18581 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
18583 * config/rs6000/rs6000.c (struct machine_function): Add field
18584 fpr_is_wrapped_separately.
18585 (rs6000_get_separate_components): Use 64 components. Handle the
18586 new FPR components.
18587 (rs6000_components_for_bb): Handle the FPR components.
18588 (rs6000_emit_prologue_components): Handle the FPR components.
18589 (rs6000_emit_epilogue_components): Handle the FPR components.
18590 (rs6000_set_handled_components): Handle the FPR components.
18591 (rs6000_emit_prologue): Don't output prologue code for those FPRs
18592 that are already separately shrink-wrapped.
18593 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
18594 that are already separately shrink-wrapped.
18596 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
18599 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
18602 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
18603 (define_peephole2 for Altivec d-form load): Add peepholes to catch
18604 cases where the register allocator uses a move and an offsettable
18605 memory operation to/from a FPR register on ISA 2.06/2.07.
18606 (define_peephole2 for Altivec d-form store): Likewise.
18608 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
18611 * config/i386/mmx.md (*mov<mode>_internal): Enable
18612 alternatives 11, 12, 13 and 14 also for 32bit targets.
18613 Remove alternatives 15, 16, 17 and 18.
18614 * config/i386/sse.md (vec_concatv2di): Change
18615 alternative (!x, *y) to (x, ?!*Yn).
18617 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
18619 * dumpfile.h (enum dump_kind): Remove stray comma.
18621 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
18623 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
18624 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
18625 predicate::num_conditions
18626 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
18627 (CHANGED): turn into predicate::changed.
18628 (agg_position_info): Move to ipa-predicate.h
18629 (add_condition, predicate::add_clause, predicate::operator &=,
18630 predicate::or_with, predicate::evaluate, predicate::probability,
18631 dump_condition, dump_clause, predicate::dump,
18632 predicate::remap_after_duplication, predicate::remap_after_inlining,
18633 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
18634 (evaluate_conditions_for_known_args): Update.
18635 (set_cond_stmt_execution_predicate): Update.
18636 * ipa-inline.h: Include ipa-predicate.h
18637 (condition, inline_param_summary, conditions, agg_position_info,
18638 predicate): Move to ipa-predicate.h
18639 * ipa-predicate.c: New file.
18640 * ipa-predicate.h: New file.
18642 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
18644 * final.c (leaf_function_p): Check we are not in a sequence.
18646 2017-05-18 Martin Liska <mliska@suse.cz>
18648 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
18649 * dumpfile.c (dump_register): Use new enum dump_kind.
18650 (get_dump_file_name): Likewise.
18651 (dump_enable_all): Likewise.
18652 (dump_switch_p_1): Likewise.
18653 (enable_rtl_dump_file): Remove usage of TDF_RTL.
18654 * dumpfile.h (enum dump_kind): New enum type.
18655 (struct dump_file_info): Create constructor and
18656 format fields and comments.
18657 * passes.c (pass_manager::register_one_dump_file):
18659 * statistics.c (statistics_early_init): Likewise.
18660 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
18661 TDF_TREE with TDF_SLIM.
18662 (gather_memory_references_ref): Likewise.
18664 2017-05-18 Martin Liska <mliska@suse.cz>
18666 * vec.h (struct vnull): Use it.
18668 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
18670 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
18671 (true_predicate, false_predicate, true_predicate_p,
18672 false_predicate_p): Remove.
18673 (single_cond_predicate, not_inlined_predicate): Turn to member function
18675 (add_condition): Update.
18676 (add_clause): Turn to...
18677 (predicate::add_clause): ... this one; update; allow passing NULL
18679 (and_predicates): Turn to ...
18680 (predicate::operator &=): ... this one.
18681 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
18682 (or_predicates): Turn to ...
18683 (predicate::or_with): ... this one.
18684 (evaluate_predicate): Turn to ...
18685 (predicate::evaluate): ... this one.
18686 (predicate_probability): Turn to ...
18687 (predicate::probability): ... this one.
18688 (dump_condition): Update.
18689 (dump_predicate): Turn to ...
18690 (predicate::dump): ... this one.
18691 (account_size_time): Update.
18692 (edge_set_predicate): Update.
18693 (set_hint_predicate): UPdate.
18694 (evaluate_conditions_for_known_args): Update.
18695 (evaluate_properties_for_edge): Update.
18696 (remap_predicate_after_duplication): Turn to...
18697 (predicate::remap_after_duplication): ... this one.
18698 (remap_hint_predicate_after_duplication): Update.
18699 (inline_summary_t::duplicate): UPdate.
18700 (dump_inline_edge_summary): Update.
18701 (dump_inline_summary): Update.
18702 (set_cond_stmt_execution_predicate): Update.
18703 (set_switch_stmt_execution_predicate): Update.
18704 (compute_bb_predicates): Update.
18705 (will_be_nonconstant_expr_predicate): Update.
18706 (will_be_nonconstant_predicate): Update.
18707 (phi_result_unknown_predicate): Update.
18708 (predicate_for_phi_result): Update.
18709 (array_index_predicate): Update.
18710 (estimate_function_body_sizes): Update.
18711 (estimate_node_size_and_time): Update.
18712 (estimate_ipcp_clone_size_and_time): Update.
18713 (remap_predicate): Rename to ...
18714 (predicate::remap_after_inlining): ... this one.
18715 (remap_hint_predicate): Update.
18716 (inline_merge_summary): Update.
18717 (inline_update_overall_summary): Update.
18718 (estimate_size_after_inlining): Update.
18719 (read_predicate): Rename to ...
18720 (predicate::stream_in): ... this one.
18721 (read_inline_edge_summary): Update.
18722 (write_predicate): Rename to ...
18723 (predicate::stream_out): ... this one.
18724 (write_inline_edge_summary): Update.
18725 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
18726 (clause_t): Turn to uint32_t
18727 (predicate): Turn to class; implement constructor and operators
18729 (size_time_entry): Update.
18730 (inline_summary): Update.
18731 (inline_edge_summary): Update.
18733 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
18735 * fold-const.c (fold_binary_loc): Move transformation...
18736 * match.pd (C - X CMP X): ... here.
18738 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
18740 * config/sparc/sparc.c (sparc_option_override): Set function
18741 alignment for -mcpu=niagara7 to 64 to match the I$ line.
18742 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
18744 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
18746 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
18748 2017-05-18 Marek Polacek <polacek@redhat.com>
18751 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
18752 (pass_ubsan::execute): Call gimple_assign_single_p instead of
18753 gimple_assign_load_p.
18755 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
18757 PR middle-end/80692
18758 * real.c (do_compare): Give decimal_do_compare preference over
18759 comparing just the signs.
18761 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
18763 * doc/md.texi (Canonicalization of Instructions): Describe the
18764 canonical form of instructions that inherently set a condition
18767 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
18769 PR middle-end/80775
18770 * tree-cfg.c: Move deletion of unreachable case statements to after
18771 the merging of consecutive case labels.
18773 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
18775 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
18776 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
18777 restoring of callee-saved registers.
18779 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
18781 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
18782 * config/visium/visium.c (single_set_and_flags): Likewise.
18783 * config/visium/visium.md (Substitutions): Likewise.
18785 2017-05-17 Martin Liska <mliska@suse.cz>
18787 * cfg.c: Introduce dump_flags_t type and
18788 use it instead of int type.
18790 * cfghooks.c: Likewise.
18791 * cfghooks.h (struct cfg_hooks): Likewise.
18792 * cfgrtl.c: Likewise.
18793 * cfgrtl.h: Likewise.
18794 * cgraph.c (cgraph_node::get_body): Likewise.
18795 * coretypes.h: Likewise.
18796 * domwalk.c: Likewise.
18797 * domwalk.h: Likewise.
18798 * dumpfile.c (struct dump_option_value_info): Likewise.
18799 (dump_enable_all): Likewise.
18800 (dump_switch_p_1): Likewise.
18801 (opt_info_switch_p): Likewise.
18802 * dumpfile.h (enum tree_dump_index): Likewise.
18803 (struct dump_file_info): Likewise.
18804 * genemit.c: Likewise.
18805 * generic-match-head.c: Likewise.
18806 * gengtype.c (open_base_files): Likewise.
18807 * gimple-pretty-print.c: Likewise.
18808 * gimple-pretty-print.h: Likewise.
18809 * graph.c (print_graph_cfg): Likewise.
18810 * graphite-scop-detection.c (dot_all_sese): Likewise.
18811 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
18812 * loop-unroll.c (report_unroll): Likewise.
18813 * passes.c (pass_manager::register_one_dump_file): Likewise.
18814 * print-tree.c: Likewise.
18815 * statistics.c: Likewise.
18816 * tree-cfg.c: Likewise.
18817 * tree-cfg.h: Likewise.
18818 * tree-dfa.c: Likewise.
18819 * tree-dfa.h: Likewise.
18820 * tree-dump.c (dump_function): Likewise.
18821 * tree-dump.h (struct dump_info): Likewise.
18822 * tree-pretty-print.c: Likewise.
18823 * tree-pretty-print.h: Likewise.
18824 * tree-ssa-live.c: Likewise.
18825 * tree-ssa-live.h: Likewise.
18826 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
18827 * tree-vect-loop.c: Likewise.
18828 * tree-vect-slp.c: Likewise.
18830 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
18831 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18833 PR tree-optimization/80457
18834 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
18835 of all arguments to a statement as scalar_to_vec operations.
18836 (vectorizable_call): Adjust call to vect_model_simple_cost for
18838 (vectorizable_conversion): Likewise.
18839 (vectorizable_assignment): Likewise.
18840 (vectorizable_shift): Likewise.
18841 (vectorizable_operation): Likewise.
18842 (vectorizable_comparison): Likewise.
18843 (vect_is_simple_cond): Record the def types for operands.
18844 (vectorizable_condition): Likewise, call vect_model_simple_cost.
18845 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
18846 for statement argument count.
18848 2017-05-16 Carl Love <cel@us.ibm.com>
18850 * config/rs6000/rs6000-c: Add support for built-in functions
18851 vector unsigned long long vec_bperm (vector unsigned long long,
18852 vector unsigned char)
18853 vector signed long long vec_mule (vector signed int,
18855 vector unsigned long long vec_mule (vector unsigned int,
18856 vector unsigned int)
18857 vector signed long long vec_mulo (vector signed int,
18859 vector unsigned long long vec_mulo (vector unsigned int,
18860 vector unsigned int)
18861 vector signed char vec_sldw (vector signed char,
18862 vector signed char,
18864 vector unsigned char vec_sldw (vector unsigned char,
18865 vector unsigned char,
18867 vector signed short vec_sldw (vector signed short,
18868 vector signed short,
18870 vector unsigned short vec_sldw (vector unsigned short,
18871 vector unsigned short,
18873 vector signed int vec_sldw (vector signed int,
18876 vector unsigned int vec_sldw (vector unsigned int,
18877 vector unsigned int,
18879 vector signed long long vec_sldw (vector signed long long,
18880 vector signed long long,
18882 vector unsigned long long vec_sldw (vector unsigned long long,
18883 vector unsigned long long,
18885 * config/rs6000/rs6000-c: Add support for built-in functions
18886 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
18887 * config/rs6000/altivec.h: Add defintion for vec_sldw.
18888 * doc/extend.texi: Update the built-in documentation for the
18889 new built-in functions.
18891 2017-05-16 Marek Polacek <polacek@redhat.com>
18895 * tree.c (save_expr): Don't fold the expression.
18897 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
18899 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
18900 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
18901 and (?*y,m). Update insn attributes.
18903 2017-05-16 Martin Liska <mliska@suse.cz>
18905 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
18906 flags argument of print_gimple_stmt, print_gimple_expr,
18907 print_generic_stmt and print_generic_expr.
18908 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
18909 * coretypes.h: Likewise.
18910 * except.c (dump_eh_tree): Likewise.
18911 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
18912 * gimple-pretty-print.h: Likewise.
18913 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
18914 (backprop::push_to_worklist): Likewise.
18915 (backprop::pop_from_worklist): Likewise.
18916 (backprop::process_use): Likewise.
18917 (backprop::intersect_uses): Likewise.
18918 (note_replacement): Likewise.
18919 * gimple-ssa-store-merging.c
18920 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
18921 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
18922 (pass_store_merging::execute): Likewise.
18923 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
18924 (ssa_base_cand_dump_callback): Likewise.
18925 (dump_incr_vec): Likewise.
18926 (replace_refs): Likewise.
18927 (replace_mult_candidate): Likewise.
18928 (create_add_on_incoming_edge): Likewise.
18929 (create_phi_basis): Likewise.
18930 (insert_initializers): Likewise.
18931 (all_phi_incrs_profitable): Likewise.
18932 (introduce_cast_before_cand): Likewise.
18933 (replace_one_candidate): Likewise.
18934 * gimplify.c (gimplify_expr): Likewise.
18935 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
18936 (set_rename): Likewise.
18937 (rename_uses): Likewise.
18938 (copy_loop_phi_nodes): Likewise.
18939 (add_close_phis_to_merge_points): Likewise.
18940 (copy_loop_close_phi_args): Likewise.
18941 (copy_cond_phi_args): Likewise.
18942 (graphite_copy_stmts_from_block): Likewise.
18943 (translate_pending_phi_nodes): Likewise.
18944 * graphite-poly.c (print_pdr): Likewise.
18945 (dump_gbb_cases): Likewise.
18946 (dump_gbb_conditions): Likewise.
18947 (print_scop_params): Likewise.
18948 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
18949 (build_cross_bb_scalars_use): Likewise.
18950 (gather_bbs::before_dom_children): Likewise.
18951 * hsa-dump.c (dump_hsa_immed): Likewise.
18952 * ipa-cp.c (print_ipcp_constant_value): Likewise.
18953 (get_replacement_map): Likewise.
18954 * ipa-inline-analysis.c (dump_condition): Likewise.
18955 (estimate_function_body_sizes): Likewise.
18956 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
18957 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
18958 * ipa-prop.c (ipa_dump_param): Likewise.
18959 (ipa_print_node_jump_functions_for_edge): Likewise.
18960 (ipa_modify_call_arguments): Likewise.
18961 (ipa_modify_expr): Likewise.
18962 (ipa_dump_param_adjustments): Likewise.
18963 (ipa_dump_agg_replacement_values): Likewise.
18964 (ipcp_modif_dom_walker::before_dom_children): Likewise.
18965 * ipa-pure-const.c (check_stmt): Likewise.
18966 (pass_nothrow::execute): Likewise.
18967 * ipa-split.c (execute_split_functions): Likewise.
18968 * omp-offload.c (dump_oacc_loop_part): Likewise.
18969 (dump_oacc_loop): Likewise.
18970 * trans-mem.c (tm_log_emit): Likewise.
18971 (tm_memopt_accumulate_memops): Likewise.
18972 (dump_tm_memopt_set): Likewise.
18973 (dump_tm_memopt_transform): Likewise.
18974 * tree-cfg.c (gimple_verify_flow_info): Likewise.
18975 (print_loop): Likewise.
18976 * tree-chkp-opt.c (chkp_print_addr): Likewise.
18977 (chkp_gather_checks_info): Likewise.
18978 (chkp_get_check_result): Likewise.
18979 (chkp_remove_check_if_pass): Likewise.
18980 (chkp_use_outer_bounds_if_possible): Likewise.
18981 (chkp_reduce_bounds_lifetime): Likewise.
18982 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
18983 (chkp_mark_completed_bounds): Likewise.
18984 (chkp_register_incomplete_bounds): Likewise.
18985 (chkp_mark_invalid_bounds): Likewise.
18986 (chkp_maybe_copy_and_register_bounds): Likewise.
18987 (chkp_build_returned_bound): Likewise.
18988 (chkp_get_bound_for_parm): Likewise.
18989 (chkp_build_bndldx): Likewise.
18990 (chkp_get_bounds_by_definition): Likewise.
18991 (chkp_generate_extern_var_bounds): Likewise.
18992 (chkp_get_bounds_for_decl_addr): Likewise.
18993 * tree-chrec.c (chrec_apply): Likewise.
18994 * tree-data-ref.c (dump_data_reference): Likewise.
18995 (dump_subscript): Likewise.
18996 (dump_data_dependence_relation): Likewise.
18997 (analyze_overlapping_iterations): Likewise.
18998 * tree-inline.c (expand_call_inline): Likewise.
18999 (tree_function_versioning): Likewise.
19000 * tree-into-ssa.c (dump_defs_stack): Likewise.
19001 (dump_currdefs): Likewise.
19002 (dump_names_replaced_by): Likewise.
19003 (dump_update_ssa): Likewise.
19004 (update_ssa): Likewise.
19005 * tree-object-size.c (pass_object_sizes::execute): Likewise.
19006 * tree-parloops.c (build_new_reduction): Likewise.
19007 (try_create_reduction_list): Likewise.
19008 (ref_conflicts_with_region): Likewise.
19009 (oacc_entry_exit_ok_1): Likewise.
19010 (oacc_entry_exit_single_gang): Likewise.
19011 * tree-pretty-print.h: Likewise.
19012 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
19013 (get_scalar_evolution): Likewise.
19014 (add_to_evolution): Likewise.
19015 (get_loop_exit_condition): Likewise.
19016 (analyze_evolution_in_loop): Likewise.
19017 (analyze_initial_condition): Likewise.
19018 (analyze_scalar_evolution): Likewise.
19019 (instantiate_scev): Likewise.
19020 (number_of_latch_executions): Likewise.
19021 (gather_chrec_stats): Likewise.
19022 (final_value_replacement_loop): Likewise.
19023 (scev_const_prop): Likewise.
19024 * tree-sra.c (dump_access): Likewise.
19025 (disqualify_candidate): Likewise.
19026 (create_access): Likewise.
19027 (reject): Likewise.
19028 (maybe_add_sra_candidate): Likewise.
19029 (create_access_replacement): Likewise.
19030 (analyze_access_subtree): Likewise.
19031 (analyze_all_variable_accesses): Likewise.
19032 (sra_modify_assign): Likewise.
19033 (initialize_constant_pool_replacements): Likewise.
19034 (find_param_candidates): Likewise.
19035 (decide_one_param_reduction): Likewise.
19036 (replace_removed_params_ssa_names): Likewise.
19037 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
19038 * tree-ssa-copy.c (dump_copy_of): Likewise.
19039 (copy_prop_visit_cond_stmt): Likewise.
19040 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
19041 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
19042 (record_equivalences_from_stmt): Likewise.
19043 * tree-ssa-dse.c (compute_trims): Likewise.
19044 (delete_dead_call): Likewise.
19045 (delete_dead_assignment): Likewise.
19046 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
19047 (forward_propagate_into_cond): Likewise.
19048 (pass_forwprop::execute): Likewise.
19049 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
19050 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
19052 (move_computations_worker): Likewise.
19053 (execute_sm): Likewise.
19054 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
19055 (remove_exits_and_undefined_stmts): Likewise.
19056 (remove_redundant_iv_tests): Likewise.
19057 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
19058 (adjust_iv_update_pos): Likewise.
19059 * tree-ssa-math-opts.c (bswap_replace): Likewise.
19060 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
19061 (value_replacement): Likewise.
19062 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
19063 * tree-ssa-pre.c (print_pre_expr): Likewise.
19064 (get_representative_for): Likewise.
19065 (create_expression_by_pieces): Likewise.
19066 (insert_into_preds_of_block): Likewise.
19067 (eliminate_insert): Likewise.
19068 (eliminate_dom_walker::before_dom_children): Likewise.
19069 (eliminate): Likewise.
19070 (remove_dead_inserted_code): Likewise.
19071 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
19072 * tree-ssa-reassoc.c (get_rank): Likewise.
19073 (eliminate_duplicate_pair): Likewise.
19074 (eliminate_plus_minus_pair): Likewise.
19075 (eliminate_not_pairs): Likewise.
19076 (undistribute_ops_list): Likewise.
19077 (eliminate_redundant_comparison): Likewise.
19078 (update_range_test): Likewise.
19079 (optimize_range_tests_var_bound): Likewise.
19080 (optimize_vec_cond_expr): Likewise.
19081 (rewrite_expr_tree): Likewise.
19082 (rewrite_expr_tree_parallel): Likewise.
19083 (linearize_expr): Likewise.
19084 (break_up_subtract): Likewise.
19085 (linearize_expr_tree): Likewise.
19086 (attempt_builtin_powi): Likewise.
19087 (attempt_builtin_copysign): Likewise.
19088 (transform_stmt_to_copy): Likewise.
19089 (transform_stmt_to_multiply): Likewise.
19090 (dump_ops_vector): Likewise.
19091 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
19092 (print_scc): Likewise.
19093 (set_ssa_val_to): Likewise.
19094 (visit_reference_op_store): Likewise.
19095 (visit_use): Likewise.
19096 (sccvn_dom_walker::before_dom_children): Likewise.
19097 (run_scc_vn): Likewise.
19098 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
19100 (expr_hash_elt::print): Likewise.
19101 (const_and_copies::pop_to_marker): Likewise.
19102 (const_and_copies::record_const_or_copy_raw): Likewise.
19103 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
19104 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
19105 (dump_predicates): Likewise.
19106 (find_uninit_use): Likewise.
19107 (warn_uninitialized_phi): Likewise.
19108 (pass_late_warn_uninitialized::execute): Likewise.
19109 * tree-ssa.c (verify_vssa): Likewise.
19110 (verify_ssa): Likewise.
19111 (maybe_optimize_var): Likewise.
19112 * tree-vrp.c (dump_value_range): Likewise.
19113 (dump_all_value_ranges): Likewise.
19114 (dump_asserts_for): Likewise.
19115 (register_edge_assert_for_2): Likewise.
19116 (vrp_visit_cond_stmt): Likewise.
19117 (vrp_visit_switch_stmt): Likewise.
19118 (vrp_visit_stmt): Likewise.
19119 (vrp_visit_phi_node): Likewise.
19120 (simplify_cond_using_ranges_1): Likewise.
19121 (fold_predicate_in): Likewise.
19122 (evrp_dom_walker::before_dom_children): Likewise.
19123 (evrp_dom_walker::push_value_range): Likewise.
19124 (evrp_dom_walker::pop_value_range): Likewise.
19125 (execute_early_vrp): Likewise.
19127 2017-05-16 Richard Biener <rguenther@suse.de>
19129 * dwarf2out.c (loc_list_from_tree_1): Do not create
19130 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
19132 2017-05-16 Richard Biener <rguenther@suse.de>
19134 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
19136 (note_variable_value_in_expr): If we resolved the decl ref
19137 do not push to the stack.
19139 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
19141 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
19142 operations in fast-math mode.
19143 (vaddq_f16): Likewise.
19144 (vmul_f16): Likewise.
19145 (vmulq_f16): Likewise.
19146 (vsub_f16): Likewise.
19147 (vsubq_f16): Likewise.
19148 * config/arm/neon.md (add<mode>3): New.
19150 (fma:<VH:mode>3): New. Also remove outdated comment.
19153 2017-05-16 Martin Liska <mliska@suse.cz>
19157 * ipa-devirt.c (warn_types_mismatch): Fix typo.
19158 (odr_types_equivalent_p): Likewise.
19160 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
19162 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
19164 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
19167 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
19168 non-interunit SSE move alternatives with '?'.
19169 (zero-extendsidi peephole2): New peephole to skip intermediate
19170 general register in SSE zero-extend sequence.
19172 2017-05-15 Jeff Law <law@redhat.com>
19174 * reorg.c (relax_delay_slots): Create a new variable to hold
19175 the temporary target rather than clobbering TARGET_LABEL.
19177 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
19178 missing argument to extract_bit_field call.
19179 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
19181 2017-05-15 Martin Liska <mliska@suse.cz>
19184 * gcc.c (process_command): Do not allow empty argument of -o option.
19186 2017-05-15 Renlin Li <renlin.li@arm.com>
19188 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
19189 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
19190 * config/aarch64/constraints.md (Usf): Add long call check.
19191 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
19192 (call_value): Likewise.
19193 (sibcall): Likewise.
19194 (sibcall_value): Likewise.
19196 (call_value_insn): New.
19197 (sibcall_insn): Update rtx pattern.
19198 (sibcall_value_insn): Likewise.
19199 (call_internal): Remove.
19200 (call_value_internal): Likewise.
19201 (sibcall_internal): Likewise.
19202 (sibcall_value_internal): Likewise.
19203 (call_reg): Likewise.
19204 (call_symbol): Likewise.
19205 (call_value_reg): Likewise.
19206 (call_value_symbol): Likewise.
19208 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
19211 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
19213 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
19215 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
19216 compatible with CCGOCmode and with CCZmode.
19218 2017-05-14 Martin Sebor <msebor@redhat.com>
19220 PR middle-end/77671
19221 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
19222 (gimple_fold_builtin_snprintf): Same.
19223 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
19224 (gimple_fold_builtin_snprintf): Same.
19225 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
19226 of character types.
19227 (is_call_safe): New function.
19228 (try_substitute_return_value): Call it.
19229 (try_simplify_call): New function.
19230 (pass_sprintf_length::handle_gimple_call): Call it.
19232 2017-05-14 Martin Sebor <msebor@redhat.com>
19234 PR middle-end/80669
19235 * builtins.c (expand_builtin_stpncpy): Simplify.
19237 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
19239 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
19240 * config/i386/i386.h
19241 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
19242 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
19243 (struct machine_function): Add new members call_ms2sysv,
19244 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
19245 (struct machine_frame_state): New fields sp_realigned and
19246 sp_realigned_offset.
19247 * config/i386/i386.c
19248 (enum xlogue_stub): New enum.
19249 (enum xlogue_stub_sets): New enum.
19250 (class xlogue_layout): New class.
19251 (struct ix86_frame): New fields stack_realign_allocate_offset,
19252 stack_realign_offset and outlined_save_offset. Modify comments to
19253 detail stack layout when using out-of-line stubs.
19254 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
19255 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
19256 -mcall-ms2sysv-xlogues.
19257 (stub_managed_regs): New static variable.
19258 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
19259 registers managed by out-of-line stub.
19260 (disable_call_ms2sysv_xlogues): New function.
19261 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
19262 m->call_ms2sysv when appropriate and compute frame layout for
19264 (sp_valid_at, fp_valid_at): New inline functions.
19265 (choose_basereg): New function.
19266 (choose_baseaddr): Add align parameter, use choose_basereg and modify
19268 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
19269 Use align parameter of choose_baseaddr to generated aligned SSE movs
19271 (pro_epilogue_adjust_stack): Modify to track
19272 machine_frame_state::sp_realigned.
19273 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
19274 (ix86_nsaved_sseregs): Likewise.
19275 (ix86_emit_save_regs): Likewise.
19276 (ix86_emit_save_regs_using_mov): Likewise.
19277 (ix86_emit_save_sse_regs_using_mov): Likewise.
19278 (get_scratch_register_on_entry): Likewise.
19279 (gen_frame_set): New function.
19280 (gen_frame_load): Likewise.
19281 (gen_frame_store): Likewise.
19282 (emit_outlined_ms2sysv_save): Likewise.
19283 (emit_outlined_ms2sysv_restore): Likewise.
19284 (ix86_expand_prologue): Modify stack re-alignment code and call
19285 emit_outlined_ms2sysv_save when appropriate.
19286 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
19287 parameter rtx_insn *insn, which allows the function to be used to only
19288 generate the notes.
19289 (ix86_expand_epilogue): Modify validity checks of frame and stack
19290 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
19291 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
19292 * config/i386/predicates.md
19293 (save_multiple): New predicate.
19294 (restore_multiple): Likewise.
19295 * config/i386/sse.md
19296 (save_multiple<mode>): New pattern.
19297 (save_multiple_realign<mode>): Likewise.
19298 (restore_multiple<mode>): Likewise.
19299 (restore_multiple_and_return<mode>): Likewise.
19300 (restore_multiple_leave_return<mode>): Likewise.
19301 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
19303 2017-05-14 Julia Koval <julia.koval@intel.com>
19305 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
19306 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
19307 (__builtin_ia32_xsetbv): New builtins.
19308 * config/i386/i386.c (ix86_expand_special_args_builtin):
19310 (ix86_expand_builtin): Special expand for new intrinsics.
19311 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
19312 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
19313 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
19315 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19317 * cfganal.c (inverted_post_order_compute): Change argument type
19319 * cfganal.h (inverted_post_order_compute): Adjust prototype.
19320 * df-core.c (rest_of_handle_df_initialize): Adjust.
19321 (rest_of_handle_df_finish): Likewise.
19322 (df_analyze_1): Likewise.
19323 (df_analyze): Likewise.
19324 (loop_inverted_post_order_compute): Change argument to be a vec *.
19325 (df_analyze_loop): Adjust.
19326 (df_get_n_blocks): Likewise.
19327 (df_get_postorder): Likewise.
19328 * df.h (struct df_d): Change field to be a vec.
19329 * lcm.c (compute_laterin): Adjust.
19330 (compute_available): Likewise.
19331 * lra-lives.c (lra_create_live_ranges_1): Likewise.
19332 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
19333 * tree-ssa-pre.c (compute_antic): Likewise.
19335 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19337 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
19338 (depth_first_search::depth_first_search): Change structure init
19339 function to this constructor.
19340 (depth_first_search::add_bb): Rename function to this member.
19341 (depth_first_search::execute): Likewise.
19342 (flow_dfs_compute_reverse_finish): Adjust.
19344 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19346 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
19347 (longest_simple_path): Likewise.
19348 * shrink-wrap.c (spread_components): Likewise.
19349 (disqualify_problematic_components): Likewise.
19350 (emit_common_heads_for_components): Likewise.
19351 (emit_common_tails_for_components): Likewise.
19352 (insert_prologue_epilogue_for_components): Likewise.
19354 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19356 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
19359 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19361 * df-core.c (df_set_blocks): Start using auto_bitmap.
19362 (df_compact_blocks): Likewise.
19363 * df-problems.c (df_rd_confluence_n): Likewise.
19364 * df-scan.c (df_insn_rescan_all): Likewise.
19365 (df_process_deferred_rescans): Likewise.
19366 (df_update_entry_block_defs): Likewise.
19367 (df_update_exit_block_uses): Likewise.
19368 (df_entry_block_bitmap_verify): Likewise.
19369 (df_exit_block_bitmap_verify): Likewise.
19370 (df_scan_verify): Likewise.
19371 * lra-constraints.c (lra_constraints): Likewise.
19372 (undo_optional_reloads): Likewise.
19373 (lra_undo_inheritance): Likewise.
19374 * lra-remat.c (calculate_gen_cands): Likewise.
19375 (do_remat): Likewise.
19376 * lra-spills.c (assign_spill_hard_regs): Likewise.
19377 (spill_pseudos): Likewise.
19378 * tree-ssa-pre.c (bitmap_set_and): Likewise.
19379 (bitmap_set_subtract_values): Likewise.
19381 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19383 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
19384 management with auto_bitmap.
19385 (fix_inter_tick): Likewise.
19386 (fix_recovery_deps): Likewise.
19387 * ira.c (add_store_equivs): Likewise.
19388 (find_moveable_pseudos): Likewise.
19389 (split_live_ranges_for_shrink_wrap): Likewise.
19390 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
19391 (rtx_reuse_manager::seen_def_p): Likewise.
19392 (rtx_reuse_manager::set_seen_def): Likewise.
19393 * print-rtl.h (class rtx_reuse_manager): Likewise.
19395 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19397 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
19399 (migrate_btr_def): Likewise.
19400 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
19401 * df-core.c (loop_post_order_compute): Likewise.
19402 (loop_inverted_post_order_compute): Likewise.
19403 * hsa-common.h: Likewise.
19404 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
19405 * init-regs.c (initialize_uninitialized_regs): Likewise.
19406 * ipa-inline.c (resolve_noninline_speculation): Likewise.
19407 (inline_small_functions): Likewise.
19408 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
19409 * ira.c (combine_and_move_insns): Likewise.
19410 (build_insn_chain): Likewise.
19411 * loop-invariant.c (find_invariants): Likewise.
19412 * lower-subreg.c (propagate_pseudo_copies): Likewise.
19413 * predict.c (tree_predict_by_opcode): Likewise.
19414 (predict_paths_leading_to): Likewise.
19415 (predict_paths_leading_to_edge): Likewise.
19416 (estimate_loops_at_level): Likewise.
19417 (estimate_loops): Likewise.
19418 * shrink-wrap.c (try_shrink_wrapping): Likewise.
19419 (spread_components): Likewise.
19420 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
19421 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
19422 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
19423 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
19424 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
19425 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
19426 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
19427 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
19428 (mark_threaded_blocks): Likewise.
19429 (thread_through_all_blocks): Likewise.
19430 * tree-ssa.c (verify_ssa): Likewise.
19431 (execute_update_addresses_taken): Likewise.
19432 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
19434 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19436 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
19438 (post_order_compute): Likewise.
19439 (inverted_post_order_compute): Likewise.
19440 (pre_and_rev_post_order_compute_fn): Likewise.
19442 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19444 * genrecog.c (int_set::int_set): Explicitly construct our
19445 auto_vec base class.
19446 * vec.h (auto_vec::auto_vec): New constructor.
19448 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19450 * bitmap.h (class auto_bitmap): New constructor taking
19451 bitmap_obstack * argument.
19453 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19455 * bitmap.h (class auto_bitmap): Change type of m_bits to
19456 bitmap_head, and adjust ctor / dtor and member operators.
19458 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
19460 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
19461 when returned register mode doesn't match original mode.
19463 2017-05-12 Jeff Law <law@redhat.com>
19464 Jakub Jelinek <jakub@redhat.com>
19466 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
19467 we look for cc setter after the compare-elim changes.
19468 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
19469 within the vector to match what compare-elim now expects.
19470 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
19471 (xorsi3_flags, one_cmplsi2_flags): Likewise.
19473 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
19474 after the compare-elim changes.
19475 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
19476 the vector to match what compare-elim now expects.
19477 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
19478 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
19479 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
19480 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
19481 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
19483 * config/visium/visium.c (single_set_and_flags): Fix where
19484 we look for cc setter after the compare-elim changes.
19485 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
19486 with the vector to match what compare-elim now expects.
19487 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
19488 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
19489 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
19490 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
19491 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
19492 (neg<mode>2_insn_set_overflow): Likewise.
19494 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
19496 PR middle-end/79794
19497 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
19498 maybe_expand_insn call, set ops[0].target. If still set after call,
19499 set alt_rtl. Add extra arg to recursive calls.
19500 (extract_bit_field): Add alt_rtl argument. Pass to
19502 * expmed.h (extract_bit_field): Fix prototype.
19503 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
19504 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
19505 to extract_bit_field_calls.
19506 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
19507 Pass alt_rtl to extract_bit_field calls.
19508 * calls.c (store_unaligned_arguments_into_psuedos)
19509 load_register_parameters): Pass extra NULL to extract_bit_field calls.
19510 * optabs.c (maybe_legitimize_operand): Clear op->target when call
19512 * optabs.h (struct expand_operand): Add target bitfield.
19514 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
19516 * compare-elim.c (try_eliminate_compare): Canonicalize
19517 operation with embedded compare to
19518 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
19519 (set (reg) (operation)].
19521 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
19523 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
19526 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
19527 cost of adding a carry flag for ADC instruction.
19528 [case MINUS]: Ignore the cost of subtracting a carry flag
19529 for SBB instruction.
19531 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
19533 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
19535 * config/rs6000/bmiintrin.h: New file.
19536 * config/rs6000/bmi2intrin.h: New file.
19537 * config/rs6000/x86intrin.h: New file.
19539 2017-05-12 Jeff Law <law@redhat.com>
19541 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
19544 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
19546 PR middle-end/80707
19547 * tree-cfg.c: Remove cfg edges of unreachable case statements.
19549 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19551 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
19552 early expansion of vector divide builtins.
19553 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
19554 builtins identified as having unsigned arguments.
19556 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19558 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
19559 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
19560 expansion of vector logical operations (and, andc, or, xor,
19563 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19565 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
19566 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
19568 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19570 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
19571 early GIMPLE expansion of vector multiplies.
19573 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19575 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
19576 TARGET_HAVE_MOVT conditional.
19577 (movt splitter): Likewise.
19579 2017-05-12 Richard Biener <rguenther@suse.de>
19581 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
19582 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
19583 Fold all stmts not inplace.
19585 2017-05-12 Richard Biener <rguenther@suse.de>
19587 PR tree-optimization/80713
19588 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
19589 inserted_exprs bit for not removed stmts.
19591 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
19593 PR middle-end/69921
19594 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
19595 parallelized" attribute for parallelized OpenACC kernels.
19596 * omp-offload.c (execute_oacc_device_lower): Use it.
19598 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
19599 Set "oacc kernels" attribute.
19600 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
19601 parameter. Adjust all users.
19602 (oacc_fn_attrib_kernels_p): Remove function.
19603 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
19604 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
19605 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
19606 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
19607 assert "oacc kernels" attribute is set.
19609 2017-05-11 Carl Love <cel@us.ibm.com>
19611 * config/rs6000/rs6000-c: Add support for built-in functions
19612 vector unsigned char vec_popcnt (vector signed char)
19613 vector unsigned char vec_popcnt (vector unsigned char)
19614 vector unsigned short vec_popcnt (vector signed short)
19615 vector unsigned short vec_popcnt (vector unsigned short)
19616 vector unsigned int vec_popcnt (vector signed int)
19617 vector unsigned int vec_popcnt (vector unsigned int)
19618 vector unsigned long long vec_popcnt (vector signed long long)
19619 vector unsigned long long vec_popcnt (vector unsigned long long)
19620 vector signed long long vec_slo (vector signed long long,
19621 vector signed char)
19622 vector signed long long vec_slo (vector signed long long,
19623 vector unsigned char)
19624 vector unsigned long long vec_slo (vector unsigned long long,
19625 vector signed char)
19626 vector unsigned long long vec_slo (vector unsigned long long,
19627 vector unsigned char)
19628 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
19629 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
19630 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
19631 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
19632 * doc/extend.texi: Update the built-in documentation file for the
19633 new built-in functions.
19635 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
19637 * attribs.h (sorted_attr_string): Move machine independent
19638 functions for target clone support from the i386 port to common
19639 code. Rename ix86_function_versions to common_function_versions.
19640 Rename make_name to make_unique_name.
19641 (common_function_versions): Likewise.
19642 (make_unique_name): Likewise.
19643 (make_dispatcher_decl): Likewise.
19644 (is_function_default_version): Likewise.
19645 * attribs.c (attr_strcmp): Likewise.
19646 (sorted_attr_string): Likewise.
19647 (common_function_versions): Likewise.
19648 (make_unique_name): Likewise.
19649 (make_dispatcher_decl): Likewise.
19650 (is_function_default_version): Likewise.
19651 * config/i386/i386.c (attr_strcmp): Likewise.
19652 (sorted_attr_string): Likewise.
19653 (ix86_function_versions): Likewise.
19654 (make_name): Likewise.
19655 (make_dispatcher_decl): Likewise.
19656 (is_function_default_version): Likewise.
19657 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
19659 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19662 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
19663 Account for direct move costs for vec_construct of integer
19666 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
19669 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
19670 (UNSPEC_STX_ATOMIC): Ditto.
19671 (loaddi_via_sse): New insn.
19672 (storedi_via_sse): Ditto.
19673 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
19674 Update corresponding peephole2 patterns.
19675 (atomic_storedi_fpu): Ditto.
19677 2017-05-11 Julia Koval <julia.koval@intel.com>
19679 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
19680 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
19682 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
19683 (__builtin_ia32_rsqrt14ss_mask): New builtins.
19684 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
19686 2017-05-11 Nathan Sidwell <nathan@acm.org>
19688 * graphite-poly.c: Include dumpfile.h.
19690 * dumpfle.h (dump_function): Declare here ...
19691 * tree-dump.h (dump_function): ... not here.
19692 * dumpfile.c: #include tree-cfg.h.
19693 (dump_function): Move here from ...
19694 * tree-dump.c (dump_function): ... here.
19695 * gimplify.c: #include splay-tree.h, not tree-dump.h.
19696 * graphite-poly.c: Don't include tree-dump.h.
19697 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
19698 * print-tree.c: Likewise.
19699 * stor-layout.c: Likewise.
19700 * tree-nested.c: Likewise.
19702 * dumpfile.c (dump_start): Use TDF_FLAGS.
19703 (dump_enable_all): Fix TDF_KIND check thinko.
19705 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
19707 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19708 array entries to represent two legal parameterizations of the
19709 overloaded __builtin_cmpb function, as represented by the
19710 P6_OV_BUILTIN_CMPB constant.
19711 (altivec_resolve_overloaded_builtin): Add special case handling
19712 for the __builtin_cmpb function, as represented by the
19713 P6_OV_BUILTIN_CMPB constant.
19714 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
19715 (BU_P6_64BIT_2): New macro.
19716 (BU_P6_OVERLOAD_2): New macro
19717 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
19718 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
19719 (CMPB): Add overload support to represent both 32-bit and 64-bit
19720 compare-bytes function.
19721 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
19722 support for TARGET_CMPB.
19723 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
19724 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
19725 documentation of the __builtin_cmpb overloaded built-in function.
19727 2017-05-11 Richard Biener <rguenther@suse.de>
19729 PR tree-optimization/80705
19730 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
19731 bases are not vectorizable.
19733 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19735 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
19736 when counting register pressure.
19738 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19740 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
19741 (iv_ca_more_deps): Renamed to ...
19742 (iv_ca_compare_deps): ... this.
19743 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
19745 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19747 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
19749 (determine_group_iv_costs): ... here.
19750 (find_inv_vars_cb): Record inv var if it's not recorded before.
19752 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19754 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
19755 (get_shiftadd_cost): Ditto.
19757 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19759 * tree-ssa-address.c: Include header file.
19760 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
19762 (add_to_parts): Refactor.
19763 (addr_to_parts): New parameter. Update use of move_hint_to_base.
19764 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
19767 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19769 PR tree-optimization/53090
19770 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
19772 (extract_cond_operands): Detect condition with IV on both sides
19773 and return COMP_IV_EXPR_2.
19774 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
19775 (rewrite_use_compare): Simplify by removing call to function
19776 extract_cond_operands.
19778 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19780 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
19781 (extract_cond_operands): Detect condition comparing against non-
19782 invariant bound and return appropriate enum value.
19783 (find_interesting_uses_cond): Update use of extract_cond_operands.
19784 Handle its return value accordingly.
19785 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
19787 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19789 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
19790 nonlinear iv_use computation in loop invariant sensitive way.
19792 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19794 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
19795 (find_iv_candidates): Call relate_compare_use_with_all_cands.
19797 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19799 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
19800 (dump_cand): Support iv_cand.inv_exprs.
19801 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
19803 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
19806 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19808 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
19810 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
19811 as local function. Include necessary header files.
19812 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
19814 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19816 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
19818 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19820 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
19821 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
19822 RSHIFT_EXPR and BIT_NOT_EXPR.
19824 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19826 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
19827 (adjust_setup_cost): New parameter supporting round up adjustment.
19828 (struct address_cost_data): Delete.
19829 (force_expr_to_var_cost): Don't bound cost with spill_cost.
19830 (split_address_cost, ptr_difference_cost): Delete.
19831 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
19832 (struct ainc_cost_data): New struct.
19833 (get_address_cost_ainc): New function.
19834 (get_address_cost, get_computation_cost): Reimplement.
19835 (determine_group_iv_cost_address): Record inv_expr for all uses of
19837 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
19838 (iv_ca_has_deps): Reimplemented to ...
19839 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
19841 (iv_ca_extend): Call iv_ca_more_deps.
19843 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19845 * tree-ssa-address.c (struct mem_address): Move to header file.
19846 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
19847 * tree-ssa-address.h (struct mem_address): Move from C file.
19848 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
19850 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19852 * tree-affine.h (aff_combination_type): New interface.
19853 (aff_combination_zero_p): Remove static.
19854 (aff_combination_const_p): New interface.
19855 (aff_combination_singleton_var_p): New interfaces.
19857 2017-05-11 Richard Biener <rguenther@suse.de>
19859 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
19860 Skip unreachable blocks and destinations.
19861 (eliminate): Move stmt removal and fixup ...
19862 (fini_eliminate): ... here. Skip inserted exprs.
19863 (pass_pre::execute): Move fini_pre after fini_eliminate.
19864 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
19865 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
19866 PRE to get rid of dead code that has invalid SSA form and
19867 split critical edges again.
19869 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19871 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19873 2017-05-11 Richard Biener <rguenther@suse.de>
19875 * passes.c (execute_function_todo): Verify loops if they are
19876 said to be up-to-date.
19877 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
19878 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
19880 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
19883 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
19884 handle calling assemble_external ourself.
19887 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
19888 modes with zero size. Enhance comment.
19890 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19892 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
19893 built-ins for vec_xl and vec_xst with short and char pointer
19896 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
19898 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
19899 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
19900 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
19901 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
19902 (_mm_maskz_min_round_ss): New intrinsics.
19903 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
19904 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
19905 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
19906 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
19907 (__builtin_ia32_minss_mask_round): New builtins.
19908 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19909 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
19910 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
19912 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
19913 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
19915 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
19918 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
19920 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
19921 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
19922 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
19923 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
19924 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
19925 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
19926 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
19927 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19928 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
19929 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
19930 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
19931 (__builtin_ia32_mulss_mask_round): New builtins.
19932 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19933 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
19934 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
19936 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
19937 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
19939 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
19942 2017-05-10 Julia Koval <julia.koval@intel.com>
19944 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
19945 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
19946 (_mm256_setr_m128i): New intrinsics.
19948 2017-05-10 Julia Koval <julia.koval@intel.com>
19950 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
19951 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
19952 (_mm_maskz_rcp14_ss): New intrinsics.
19953 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
19954 (__builtin_ia32_rcp14ss_mask): New builtins.
19955 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
19957 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
19959 PR tree-optimization/51513
19960 * tree-cfg.c (gimple_seq_unreachable_p): New function.
19961 (assert_unreachable_fallthru_edge_p): Use it.
19962 (group_case_labels_stmt): Likewise.
19963 * tree-cfg.h: Prototype it.
19964 * stmt.c: Include cfghooks.h and tree-cfg.h.
19965 (emit_case_dispatch_table) <gap_label>: New local variable.
19966 Use it to fill dispatch table gaps.
19967 Test for default_label before updating probabilities.
19968 (expand_case) <default_label>: Remove unneeded initialization.
19969 Test for unreachable default case statement and remove its edge.
19970 Set default_label accordingly.
19971 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
19973 2017-05-10 Carl Love <cel@us.ibm.com>
19975 * config/rs6000/rs6000-c: Add support for built-in functions
19976 vector signed char vec_neg (vector signed char)
19977 vector signed short int vec_neg (vector short int)
19978 vector signed int vec_neg (vector signed int)
19979 vector signed long long vec_neg (vector signed long long)
19980 vector float vec_neg (vector float)
19981 vector double vec_neg (vector double)
19982 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
19984 * config/rs6000/altivec.h: Add define for vec_neg
19985 * doc/extend.texi: Update the built-in documentation for the
19986 new built-in functions.
19988 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19990 PR tree-optimization/77644
19991 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
19993 2017-05-10 Nathan Sidwell <nathan@acm.org>
19995 * dumpfile.h (TDI_lang_all): New.
19996 (TDF_KIND): New. Renumber others
19997 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
19999 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
20001 (get_dump_file_name): Adjust suffix generation.
20002 (dump_enable_all): Use TDF_KIND.
20003 * doc/invoke.texi (-fdump-lang-all): Document.
20005 * dumpfile.h: Tabify.
20007 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
20010 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
20011 Move member access before delete.
20013 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
20015 * tree-inline.c (expand_call_inline): Split block at stmt
20018 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
20021 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
20022 are now unused after splitting mov{sf,sd}_hardfloat.
20023 (f32_lr2): Likewise.
20024 (f32_lm): Likewise.
20025 (f32_lm2): Likewise.
20026 (f32_li): Likewise.
20027 (f32_li2): Likewise.
20028 (f32_lv): Likewise.
20029 (f32_sr): Likewise.
20030 (f32_sr2): Likewise.
20031 (f32_sm): Likewise.
20032 (f32_sm2): Likewise.
20033 (f32_si): Likewise.
20034 (f32_si2): Likewise.
20035 (f32_sv): Likewise.
20036 (f32_dm): Likewise.
20037 (f32_vsx): Likewise.
20038 (f32_av): Likewise.
20039 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
20040 For movsf, order stores so the VSX stores occur before the GPR
20041 store which encourages the register allocator to use a traditional
20042 FPR instead of a GPR. For movsd, order the stores so that the GPR
20043 store comes before the VSX stores to allow the power6 to work.
20044 This is due to the power6 not having a 32-bit integer store
20045 instruction from a FPR.
20046 (movsf_hardfloat): Likewise.
20047 (movsd_hardfloat): Likewise.
20049 2017-05-09 Martin Sebor <msebor@redhat.com>
20051 PR translation/80280
20052 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
20055 PR translation/80280
20056 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
20057 data member added in r247778.
20058 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
20060 2017-05-09 Nathan Sidwell <nathan@acm.org>
20062 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
20064 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
20067 2017-05-09 Marek Polacek <polacek@redhat.com>
20069 * doc/invoke.texi: Fix typo.
20071 2017-05-09 Richard Biener <rguenther@suse.de>
20073 * tree-vrp.c (vrp_val_is_max): Adjust comment.
20074 (vrp_val_is_min): Likewise.
20075 (set_value_range_to_value): Likewise.
20076 (set_value_range_to_nonnegative): Likewise.
20077 (gimple_assign_nonzero_p): Likewise.
20078 (gimple_stmt_nonzero_p): Likewise.
20079 (vrp_int_const_binop): Likewise. Remove unreachable case.
20080 (adjust_range_with_scev): Adjust comments.
20081 (compare_range_with_value): Likewise.
20082 (extract_range_from_phi_node): Likewise.
20083 (test_for_singularity): Likewise.
20085 2017-05-09 Richard Biener <rguenther@suse.de>
20087 * tree-vrp.c (get_single_symbol): Add assert that we don't
20088 get overflowed constants as invariant part.
20089 (compare_values_warnv): Add comment before the TREE_NO_WARNING
20090 checks. Use wi::cmp instead of recursing for integer constants.
20091 (compare_values): Just ignore whether we assumed undefined
20092 overflow instead of failing the compare.
20093 (extract_range_for_var_from_comparison_expr): Add comment before the
20094 TREE_NO_WARNING sets.
20095 (test_for_singularity): Likewise.
20096 (extract_range_from_comparison): Do not disable optimization
20097 when we assumed undefined overflow.
20098 (extract_range_basic): Remove init of unused var.
20100 2017-05-09 Richard Biener <rguenther@suse.de>
20102 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
20103 (extract_range_from_multiplicative_op_1): Adjust.
20104 (extract_range_from_binary_expr_1): Use int_const_binop.
20106 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
20109 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
20110 rs6000_store_data_bypass_p in seven define_bypass directives and
20111 in several comments.
20112 * config/rs6000/rs6000-protos.h: Add prototype for
20113 rs6000_store_data_bypass_p function.
20114 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
20115 function implements slightly different (rs6000-specific) semantics
20116 than store_data_bypass_p, returning false rather than aborting
20117 with assertion error when arguments do not satisfy the
20118 requirements of store data bypass.
20119 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
20120 rs6000_store_data_bypass_p.
20122 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
20124 * config/xtensa/xtensa-protos.h
20125 (xtensa_initial_elimination_offset): New declaration.
20126 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
20127 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
20128 macro definition, add case for FRAME_POINTER_REGNUM when
20129 FRAME_GROWS_DOWNWARD.
20130 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
20131 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
20132 xtensa_initial_elimination_offset.
20134 2017-05-08 Nathan Sidwell <nathan@acm.org>
20136 * doc/invoke.texi: Alphabetize -fdump options.
20138 2017-05-08 Martin Sebor <msebor@redhat.com>
20140 PR translation/80280
20141 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
20143 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
20145 * target.def (compute_frame_layout): New optional target hook.
20146 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
20147 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
20148 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
20150 * reload1.c (verify_initial_elim_offsets): Likewise.
20151 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
20152 (use_simple_return_p): Call arm_compute_frame_layout if needed.
20153 (arm_get_frame_offsets): Split up into this ...
20154 (arm_compute_frame_layout): ... and this function.
20156 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
20158 * config/aarch64/constraints.md (Usa): New constraint.
20159 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
20161 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
20163 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
20164 with_multilib_list after it has been checked.
20166 2017-05-08 Richard Biener <rguenther@suse.de>
20168 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
20169 (bitmap_set_subtract_values): Likewise.
20171 2017-05-08 Richard Biener <rguenther@suse.de>
20173 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
20174 (gimple_assign_nonzero): ... this and remove strict_overflow_p
20176 (gimple_stmt_nonzero_warnv_p): Rename to ...
20177 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
20179 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
20180 (extract_range_basic): Adjust, do not disable propagation on
20181 strict overflow sensitive simplification.
20182 (vrp_visit_cond_stmt): Likewise.
20184 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
20186 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
20187 body size unconditionally.
20189 2017-05-07 Jeff Law <law@redhat.com>
20192 2017-05-06 Jeff Law <law@redhat.com>
20193 PR tree-optimization/78496
20194 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
20197 PR tree-optimization/78496
20198 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
20199 (simplify_stmt_using_ranges): Call it.
20200 (vrp_dom_walker::before_dom_children): Extract equivalences
20201 from an ASSERT_EXPR with an equality comparison against a
20204 2017-05-06 Jeff Law <law@redhat.com>
20206 PR tree-optimization/78496
20207 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
20210 PR tree-optimization/78496
20211 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
20212 (simplify_stmt_using_ranges): Call it.
20213 (vrp_dom_walker::before_dom_children): Extract equivalences
20214 from an ASSERT_EXPR with an equality comparison against a
20217 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
20219 * lra-constraints.c (lra_copy_reg_equiv): New function.
20220 (split_reg): Use it to copy equivalence information from the
20221 original register to the spill register.
20223 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
20225 PR rtl-optimization/75964
20226 * simplify-rtx.c (simplify_const_relational_operation): Remove
20227 invalid handling of comparisons of integer ABS.
20229 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
20231 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
20232 initialize to zero.
20233 (init_regs): Remove declaration.
20234 (function_arg_advance_32): Initialize error_p as boolean variable.
20236 2017-05-05 Nathan Sidwell <nathan@acm.org>
20238 * store-motion.c (remove_reachable_equiv_notes): Reformat long
20239 lines. Use for (;;).
20241 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20243 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
20244 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
20245 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
20246 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
20247 VF=2 that require versioning.
20249 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20251 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
20254 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20256 * diagnostic.h (diagnostic_override_option_index): Convert from
20257 macro to inline function.
20259 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20261 * diagnostic.c (last_module_changed_p): New function.
20262 (set_last_module): New function.
20263 (diagnostic_report_current_module): Convert macro usage to
20264 the above functions.
20265 * diagnostic.h (diagnostic_context::last_module): Strengthen
20266 from const line_map * to const line_map_ordinary *.
20267 (diagnostic_last_module_changed): Delete macro.
20268 (diagnostic_set_last_module): Delete macro.
20270 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20272 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
20273 with diagnostic_report_diagnostic.
20274 (diagnostic_n_impl_richloc): Likewise.
20275 * diagnostic.h (report_diagnostic): Delete macro.
20276 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
20277 with diagnostic_report_diagnostic.
20278 * substring-locations.c (format_warning_va): Likewise.
20280 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20282 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
20283 save/restor of format_spec. Move option-printing code to...
20284 (print_option_information): ...this new function, and
20285 reimplement by simply printing to the pretty_printer,
20286 rather than appending to the format string.
20288 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20290 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
20291 handling logic into...
20292 (update_effective_level_from_pragmas): ...this new function.
20294 2017-05-04 Andrew Waterman <andrew@sifive.com>
20296 * config/riscv/riscv.opt (mstrict-align): New option.
20297 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
20298 (SLOW_UNALIGNED_ACCESS): Define.
20299 (riscv_slow_unaligned_access): Declare.
20300 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
20302 (riscv_slow_unaligned_access): New variable.
20303 (rocket_tune_info): Set slow_unaligned_access to true.
20304 (optimize_size_tune_info): Set slow_unaligned_access to false.
20305 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
20306 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
20307 (riscv_option_override): Set riscv_slow_unaligned_access.
20308 * doc/invoke.texi: Add -mstrict-align to RISC-V.
20310 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
20312 * config/riscv/riscv.md: Unify indentation.
20314 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20319 * config/rs6000/rs6000.md (u code attribute): Add FIX and
20321 (extendsi<mode>2): Add support for doing sign extension via
20322 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
20323 don't have ISA 3.0 instructions.
20324 (extendsi<mode>2 splitter): Likewise.
20325 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
20326 generate the normal insns since SImode can now go in vector
20327 registers. Disallow the special UNSPECs needed for previous
20328 machines to hide SImode being used. Add new insns
20329 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
20330 (fix_trunc<mode>si2_stfiwx): Likewise.
20331 (fix_trunc<mode>si2_internal): Likewise.
20332 (fixuns_trunc<mode>si2): Likewise.
20333 (fixuns_trunc<mode>si2_stfiwx): Likewise.
20334 (fctiw<u>z_<mode>_smallint): Likewise.
20335 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
20336 of floating point to 32-bit integer from doing a direct move to
20337 the GPR registers to do a store.
20338 (fctiwz_<mode>): Break long line.
20340 2017-05-05 Bin Cheng <bin.cheng@arm.com>
20342 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
20343 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
20344 (addr_list, addr_offset_valid_p): New.
20345 (split_address_groups): Check offset validity with above function.
20346 (gt-tree-ssa-loop-ivopts.h): Include header file.
20348 2017-05-05 Nathan Sidwell <nathan@acm.org>
20350 * config.gcc (arm*-*-*): Add missing 'fi'.
20352 2017-05-05 Steve Ellcey <sellcey@cavium.com>
20354 * doc/invoke.texi (-fopt-info): Explicitly say order of options
20355 included in -fopt-info does not matter.
20356 * doc/optinfo.texi (-fopt-info): Fix description of default
20357 behavour. Explicitly say order of options included in -fopt-info
20360 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
20362 * config.gcc: Allow combinations of aprofile and rmprofile values for
20363 --with-multilib-list.
20364 * config/arm/t-multilib: New file.
20365 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
20366 variables. Remove setting of ISA and floating-point ABI in
20367 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
20368 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
20369 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
20370 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
20372 * config/arm/t-rmprofile: Likewise except for the matches changes.
20373 * doc/install.texi (--with-multilib-list): Document the combination of
20374 aprofile and rmprofile values and warn about pitfalls in doing that.
20376 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
20378 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
20379 (movdi_aarch64): Likewise.
20381 2017-05-05 Jakub Jelinek <jakub@redhat.com>
20383 PR tree-optimization/80632
20384 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
20386 (build_arrays): Initialize it for virtual phis.
20387 (fix_phi_nodes): Use it for virtual phis.
20389 PR tree-optimization/80558
20390 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
20391 [x, y] op z into [x op, y op z] for op & or | if conditions
20394 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
20395 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20398 * config/arm/arm.md (use_literal_pool): Remove.
20399 (64-bit immediate split): No longer takes cost into consideration
20400 if arm_disable_literal_pool is enabled.
20401 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
20402 used when arm_disable_literal_pool is enabled.
20403 (arm_max_const_double_inline_cost): Remove use of
20404 arm_disable_literal_pool.
20405 (push_minipool_fix): Add assert.
20406 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
20407 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
20408 (no_literal_pool_sf_immediate): New.
20410 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20412 PR tree-optimization/80613
20413 * tree-ssa-dce.c (propagate_necessity): Remove cases for
20414 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
20416 2017-05-05 Richard Biener <rguenther@suse.de>
20418 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
20420 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
20422 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
20423 of this flag from insn conditions due to removal from r247495.
20425 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
20427 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
20429 (arm_early_store_addr_dep_ptr): Likewise.
20430 * config/arm/aarch-common-protos.h
20431 (arm_early_load_addr_dep_ptr): Add prototype.
20432 (arm_early_store_addr_dep_ptr): Likewise.
20433 * config/arm/cortex-a53.md: Add new bypasses.
20435 2017-05-05 Jakub Jelinek <jakub@redhat.com>
20437 * tree.c (next_type_uid): Change type to unsigned.
20438 (type_hash_canon): Decrement back next_type_uid if
20439 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
20440 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
20443 2017-05-04 Martin Sebor <msebor@redhat.com>
20445 * builtins.c: Fix a trivial typo in a comment.
20447 PR middle-end/79234
20448 * builtins.c (check_sizes): Adjust to handle reading past the end.
20449 Avoid printing excessive upper bound of ranges. Use %E to print
20450 tree nodes instead of converting them to %wu.
20451 (expand_builtin_memchr): New function.
20452 (compute_dest_size): Rename...
20453 (compute_objsize): ...to this.
20454 (expand_builtin_memcpy): Adjust.
20455 (expand_builtin_mempcpy): Adjust.
20456 (expand_builtin_strcat): Adjust.
20457 (expand_builtin_strcpy): Adjust.
20458 (check_strncat_sizes): Adjust.
20459 (expand_builtin_strncat): Adjust.
20460 (expand_builtin_strncpy): Adjust and simplify.
20461 (expand_builtin_memset): Adjust.
20462 (expand_builtin_bzero): Adjust.
20463 (expand_builtin_memcmp): Adjust.
20464 (expand_builtin): Handle memcmp.
20465 (maybe_emit_chk_warning): Check strncat just once.
20467 2017-05-04 Martin Sebor <msebor@redhat.com>
20469 PR preprocessor/79214
20470 PR middle-end/79222
20471 PR middle-end/79223
20472 * builtins.c (check_sizes): Add inlining context and issue
20473 warnings even when -Wno-system-headers is set.
20474 (check_strncat_sizes): Same.
20475 (expand_builtin_strncat): Same.
20476 (expand_builtin_memmove): New function.
20477 (expand_builtin_stpncpy): Same.
20478 (expand_builtin): Handle memmove and stpncpy.
20480 2017-05-04 Bin Cheng <bin.cheng@arm.com>
20482 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
20483 which is not used any more.
20485 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20487 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
20489 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20491 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
20492 (cortexa53_tunings): Likewise.
20493 (cortexa57_tunings): Likewise.
20494 (cortexa72_tunings): Likewise.
20495 (cortexa73_tunings): Likewise.
20497 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20499 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
20500 Set loop alignment to 8.
20502 2017-05-04 Martin Sebor <msebor@redhat.com>
20504 PR translation/80280
20505 * builtins.c (expand_builtin_object_size): Add missing quoting to
20506 %D and like directives.
20507 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
20508 (hsa_type_for_tree_type): Same.
20509 (verify_function_arguments): Same.
20510 * symtab.c (symbol_table::change_decl_assembler_name): Same.
20511 * varasm.c (get_section): Same.
20514 2017-05-04 Martin Sebor <msebor@redhat.com>
20516 PR translation/80280
20517 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
20519 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20521 * config/aarch64/aarch64.c (generic_addrcost_table):
20522 Change HI/TI mode setting.
20524 2017-05-04 Martin Jambor <mjambor@suse.cz>
20526 PR tree-optimization/80622
20527 * tree-sra.c (comes_initialized_p): New function.
20528 (build_accesses_from_assign): Only set write lazily when
20529 comes_initialized_p is false.
20530 (analyze_access_subtree): Use comes_initialized_p.
20531 (propagate_subaccesses_across_link): Assert !comes_initialized_p
20532 instead of testing for PARM_DECL.
20534 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20536 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
20537 constraint on operand 0 to allow more general addressing modes.
20538 Adjust output template.
20539 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
20541 * config/aarch64/aarch64-protos.h
20542 (aarch64_address_valid_for_prefetch_p): Declare prototype.
20543 * config/aarch64/constraints.md (Dp): New address constraint.
20544 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
20547 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
20549 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
20550 update use of estimate_ipcp_clone_size_and_time.
20551 (estimate_local_effects): Update use of
20552 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
20553 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
20554 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
20555 Return nonspecialized time.
20557 2017-05-04 Richard Biener <rguenther@suse.de>
20559 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
20560 for the last VUSE which def dominates the PHI. Directly call
20562 (get_continuation_for_phi_1): Remove.
20564 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
20566 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
20567 to explain the use of truncating division. Cap the number of
20568 iterations to the maximum given by nb_iterations_upper_bound,
20571 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20573 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
20574 * configure: Regenerate.
20575 * config.in: Regenerate.
20576 * config/i386/driver-mingw32.c: new file.
20577 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
20578 * config.host: Link driver-mingw32.o on MinGW host.
20579 * doc/install.texi: Document new --enable-mingw-wildcard configure
20582 2017-05-04 Marek Polacek <polacek@redhat.com>
20584 PR tree-optimization/80612
20585 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
20587 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20588 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
20590 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
20591 (movt splitter): Likewise.
20592 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
20593 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
20594 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
20595 block for Thumb-1 with MOVT.
20596 (thumb2_legitimate_address_p): Move code block ...
20597 (can_avoid_literal_pool_for_label_p): ... into this new function.
20598 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
20600 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
20601 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
20602 "M-profile targets with the MOVT instruction".
20604 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20606 * config/arm/arm-builtins.c (arm_init_builtins): Rename
20607 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
20608 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
20610 2017-05-04 Martin Liska <mliska@suse.cz>
20612 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
20613 variable cond_code.
20615 2017-05-04 Richard Biener <rguenther@suse.de>
20617 * tree.c (array_at_struct_end_p): Handle arrays at struct
20618 end with flexarrays more conservatively. Refactor and treat
20619 arrays of arrays or aggregates more strict. Fix
20620 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
20621 * tree.c (array_at_struct_end_p): Adjust prototype.
20622 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
20623 * gimple-fold.c (get_range_strlen): Likewise.
20624 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
20626 2017-05-04 Richard Biener <rguenther@suse.de>
20628 PR tree-optimization/31130
20629 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
20631 (supports_overflow_infinity): Likewise.
20632 (is_negative_overflow_infinity): Likewise.
20633 (is_positive_overflow_infinity): Likewise.
20634 (is_overflow_infinity): Likewise.
20635 (stmt_overflow_infinity): Likewise.
20636 (overflow_infinity_range_p): Likewise.
20637 (usable_range_p): Remove as always returning true.
20638 (make_overflow_infinity): Remove.
20639 (negative_overflow_infinity): Likewise.
20640 (positive_overflow_infinity): Likewise.
20641 (avoid_overflow_infinity): Likewise.
20642 (set_value_range): Adjust accordingly.
20643 (set_value_range_to_nonnegative): Likewise, remove now unused
20644 overflow_infinity arg.
20645 (vrp_operand_equal_p): Adjust.
20646 (update_value_range): Likewise.
20647 (range_int_cst_singleton_p): Likewise.
20648 (operand_less_p): Likewise.
20649 (compare_values_warnv): Likewise.
20650 (extract_range_for_var_from_comparison_expr): Likewise.
20651 (vrp_int_const_binop): Likewise.
20652 (zero_nonzero_bits_from_vr): Likewise.
20653 (extract_range_from_multiplicative_op_1): Likewise.
20654 (extract_range_from_binary_expr_1): Likewise.
20655 (extract_range_from_unary_expr): Likewise.
20656 (extract_range_from_comparison): Likewise.
20657 (extract_range_basic): Likewise.
20658 (adjust_range_with_scev): Likewise.
20659 (compare_ranges): Likewise.
20660 (compare_range_with_value): Likewise.
20661 (dump_value_range): Likewise.
20662 (test_for_singularity): Likewise, remove strict_overflow_p parameter
20664 (simplify_cond_using_ranges): Adjust.
20666 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
20668 * brig-builtins.def: Added a builtin for class_f64.
20669 * builtin-types.def: Added a builtin type needed by class_f64.
20671 2017-05-03 Jason Merrill <jason@redhat.com>
20673 * timevar.def: Add TV_CONSTEXPR.
20675 2017-05-03 David Malcolm <dmalcolm@redhat.com>
20677 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
20679 2017-05-03 Martin Jambor <mjambor@suse.cz>
20681 * ipa-prop.c (ipa_update_after_lto_read): Removed.
20682 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
20683 * ipa-cp.c (ipcp_propagate_stage): Do not call
20684 ipa_update_after_lto_read.
20685 * ipa-inline.c (ipa_inline): Likewise.
20687 2017-05-03 Martin Jambor <mjambor@suse.cz>
20689 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
20690 tag. Added a default constructor and a destructor.
20691 (ipa_edge_args_sum_t): New class;
20692 (ipa_edge_args_sum): Declare.
20693 (ipa_edge_args_vector): Remove declaration.
20694 (IPA_EDGE_REF): Use ipa_edge_args_sum.
20695 (ipa_free_edge_args_substructures): Remove declaration.
20696 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
20697 (ipa_edge_args_info_available_for_edge_p): Likewise.
20698 * ipa-prop.c (ipa_edge_args_vector): Removed.
20699 (edge_removal_hook_holder): Likewise.
20700 (edge_duplication_hook_holder): Likewise.
20701 (ipa_edge_args_sum): New variable.
20702 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
20703 ipa_edge_args_vector.
20704 (ipa_free_edge_args_substructures): Likewise.
20705 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
20706 ipa_edge_args_vector.
20707 (ipa_edge_removal_hook): Turned into method
20708 ipa_edge_args_sum_t::remove.
20709 (ipa_edge_duplication_hook): Turned into method
20710 ipa_edge_args_sum_t::duplicate.
20711 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
20712 registering edge hooks.
20713 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
20714 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
20715 ipa_edge_args_sum instead of ipa_edge_args_vector.
20716 * ipa-profile.c (ipa_profile): Likewise.
20718 2017-05-03 Martin Jambor <mjambor@suse.cz>
20720 * symbol-summary.h (function_summary): New method exists.
20721 (function_summary::symtab_removal): Deallocate through release.
20722 (call_summary): New class.
20723 (gt_ggc_mx): New overload.
20724 (gt_pch_nx): Likewise.
20725 (gt_pch_nx): Likewise.
20727 2017-05-03 Jeff Law <law@redhat.com>
20729 PR tree-optimization/78496
20730 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
20731 from simplify_cond_using_ranges. Split off code to walk
20732 backwards through casts into ...
20733 (simplify_cond_using_ranges_2): New function.
20734 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
20735 (execute_vrp): After identifying jump threads, call
20736 simplify_cond_using_ranges_2.
20738 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
20741 * ipa-inline.h (inline_summary): Add ctor.
20742 (create_ggc): Do not use ggc_cleared_alloc.
20744 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
20745 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20747 * gcc.c (handle_braces): Support escaping in switch matching
20749 * doc/invoke.texi (Spec Files): Document it.
20750 Remove superfluous @code markup in items.
20752 2017-05-03 David Malcolm <dmalcolm@redhat.com>
20754 * diagnostic-show-locus.c (struct column_range): New struct.
20755 (get_affected_columns): New function.
20756 (get_printed_columns): New function.
20757 (struct correction): New struct.
20758 (correction::ensure_capacity): New function.
20759 (correction::ensure_terminated): New function.
20760 (struct line_corrections): New struct.
20761 (line_corrections::~line_corrections): New dtor.
20762 (line_corrections::add_hint): New function.
20763 (layout::print_trailing_fixits): Reimplement in terms of the new
20765 (selftest::test_overlapped_fixit_printing): New function.
20766 (selftest::diagnostic_show_locus_c_tests): Call it.
20768 2017-05-03 Nathan Sidwell <nathan@acm.org>
20770 Canonicalize canonical type hashing
20771 * tree.h (type_hash_canon_hash): Declare.
20772 * tree.c (type_hash_list, attribute_hash_list): Move into
20773 type_hash_canon_hash.
20774 (build_type_attribute_qual_variant): Break out hash code calc into
20775 type_hash_canon_hash.
20776 (type_hash_canon_hash): New. Generic type hash computation.
20777 (build_range_type_1, build_array_type_1, build_function_type,
20778 build_method_type_directly, build_offset_type, build_complex_type,
20779 make_vector_type): Call it.
20781 2017-05-03 Richard Biener <rguenther@suse.de>
20783 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
20784 When all DRs have unknown misaligned do not always peel
20785 when there is a store but apply the same costing model as if
20786 there were only loads.
20788 2017-05-03 Richard Biener <rguenther@suse.de>
20791 PR tree-optimization/80492
20792 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
20793 compare_base_decls returning dont-know properly.
20795 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
20797 * config/arm/iterators.md (CCSI): New mode iterator.
20798 (arch): New mode attribute.
20799 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
20800 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
20801 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
20802 code iterator for success result mode.
20803 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
20804 the corresponding new insn generators.
20806 2017-05-03 Bin Cheng <bin.cheng@arm.com>
20809 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20810 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
20812 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
20814 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
20815 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
20816 (DDR_A): Wrap DDR argument in brackets.
20817 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
20818 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
20819 (DDR_REVERSED_P): Likewise.
20821 2017-05-03 Jakub Jelinek <jakub@redhat.com>
20823 PR tree-optimization/79472
20824 * tree-switch-conversion.c (struct switch_conv_info): Add
20825 contiguous_range and default_case_nonstandard fields.
20826 (collect_switch_conv_info): Compute contiguous_range and
20827 default_case_nonstandard fields, don't clear final_bb if
20828 contiguous_range and only the default case doesn't have the required
20830 (check_all_empty_except_final): Set default_case_nonstandard instead
20831 of failing if contiguous_range and the default case doesn't have empty
20833 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
20834 and only the default case doesn't have the required constants. Skip
20836 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
20837 if default_case_nonstandard.
20838 (build_constructors): Build constant 1 just once. Assert that default
20839 values aren't inserted in between cases if contiguous_range. Skip
20841 (build_arrays): Skip virtual phis.
20842 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
20843 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
20844 Handle virtual phis.
20845 (gen_inbound_check): Handle default_case_nonstandard case.
20846 (process_switch): Adjust check_final_bb caller. Call
20847 gather_default_values with the first non-default case instead of
20848 default case if default_case_nonstandard.
20850 2017-05-02 Nathan Sidwell <nathan@acm.org>
20852 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
20853 check. Fix formatting.
20855 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
20857 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
20858 errors when comparing specialized and unspecialized times.
20860 2017-05-02 David Malcolm <dmalcolm@redhat.com>
20862 * diagnostic-show-locus.c
20863 (layout::should_print_annotation_line_p): Make private.
20864 (layout::print_annotation_line): Make private.
20865 (layout::annotation_line_showed_range_p): Make private.
20866 (layout::show_ruler): Make private.
20867 (layout::print_source_line): Make private. Pass in line and
20868 line_width, rather than calling location_get_source_line. Drop
20870 (layout::print_leading_fixits): New method.
20871 (layout::print_any_fixits): Rename to...
20872 (layout::print_trailing_fixits): ...this, and make private.
20873 Don't print newline fixits.
20874 (diagnostic_show_locus): Move logic for printing one row into...
20875 (layout::print_line): ...this new function. Move the
20876 location_get_source_line call and error-handling from
20877 print_source_line to here. Call print_leading_fixits, and rename
20878 print_any_fixits to print_trailing_fixits.
20879 (selftest::test_fixit_insert_containing_newline): Update now that
20880 newlines are partially supported.
20881 (selftest::test_fixit_insert_containing_newline_2): New test.
20882 (selftest::test_fixit_replace_containing_newline): Update comments.
20883 (selftest::diagnostic_show_locus_c_tests): Call the new test.
20884 * edit-context.c (class added_line): New class.
20885 (class edited_line): Describe newline handling in comment.
20886 (edited_line::actually_edited_p): New method.
20887 (edited_line::print_content): Delete redundant decl.
20888 (edited_line::m_predecessors): New field.
20889 (edited_file::print_content): Call edited_line::print_content.
20890 (edited_file::print_diff): Update to support newlines.
20891 (edited_file::print_diff_hunk): Likewise.
20892 (edited_file::print_run_of_changed_lines): New function.
20893 (edited_file::print_diff_line): Convert to...
20894 (print_diff_line): ...this.
20895 (edited_file::get_effective_line_count): New function.
20896 (edited_line::edited_line): Initialize new field m_predecessors.
20897 (edited_line::~edited_line): Clean up m_predecessors.
20898 (edited_line::apply_fixit): Handle newlines.
20899 (edited_line::get_effective_line_count): New function.
20900 (edited_line::print_content): New function.
20901 (edited_line::print_diff_lines): New function.
20902 (selftest::test_applying_fixits_insert_containing_newline): New
20904 (selftest::test_applying_fixits_replace_containing_newline): New
20906 (selftest::insert_line): New function.
20907 (selftest::test_applying_fixits_multiple_lines): Add example of
20909 (selftest::edit_context_c_tests): Call the new tests.
20911 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20913 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
20914 parameter cand. Update dump information.
20915 (get_computation_cost): Update uses.
20917 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20919 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
20920 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
20921 (get_computation_at, rewrite_use_address): Update use of
20922 get_computation_aff.
20924 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20926 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
20927 (get_computation): Delete.
20928 (get_computation_cost): Implement like get_computation_cost_at.
20929 Use get_computation_at.
20930 (get_computation_cost_at): Delete.
20931 (rewrite_use_nonlinear_expr): Use get_computation_at.
20932 (rewrite_use_compare, remove_unused_ivs): Ditto.
20934 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20936 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
20938 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20940 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
20941 (ivopts_global_cost_for_size): Rename parameter and update uses.
20942 (iv_ca_recount_cost): Update uses.
20943 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
20944 candidates seperately in n_invs and n_cands.
20945 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
20947 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20949 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
20950 (find_inv_vars_cb): New.
20951 (find_depends): Renamed to ...
20952 (find_inv_vars): ... this.
20953 (add_candidate_1, force_var_cost): Call find_inv_vars.
20954 (split_address_cost, determine_group_iv_cost_cond): Ditto.
20956 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20958 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
20959 inv_vars. Add inv_exprs.
20960 (struct iv_cand): Rename depends_on to inv_vars.
20961 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
20962 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
20963 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
20964 (dump_cand): Dump inv_vars.
20965 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
20966 (record_invariant, find_depends, add_candidate_1): Ditto.
20967 (set_group_iv_cost, force_var_cost): Ditto.
20968 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
20969 (get_computation_cost_at, get_computation_cost): Ditto.
20970 (determine_group_iv_cost_generic): Ditto.
20971 (determine_group_iv_cost_address): Ditto.
20972 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
20973 (determine_group_iv_costs): Ditto.
20974 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
20975 (iv_ca_set_remove_invariants): Renamed to ...
20976 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
20977 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
20978 (iv_ca_set_add_invariants): Renamed to ...
20979 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
20980 (iv_ca_set_cp): Use iv_ca_set_add_invs.
20981 (iv_ca_has_deps): Support inv_vars and inv_exprs.
20982 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
20983 (create_new_ivs): Remove useless dump.
20985 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20987 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
20989 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
20990 (iv_ca_set_no_cp, create_new_iv): Ditto.
20992 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20994 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
20996 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
20998 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
20999 function tree_check2.
21001 2017-05-02 Martin Liska <mliska@suse.cz>
21003 * doc/gcov.texi: Add missing preposition.
21004 * gcov.c (function_info::function_info): Properly fill up
21005 all member variables.
21007 2017-05-02 Tamar Christina <tamar.christina@arm.com>
21009 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
21011 2017-05-02 Tamar Christina <tamar.christina@arm.com>
21013 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
21015 2017-05-02 Martin Liska <mliska@suse.cz>
21018 * lto-streamer-in.c (lto_read_tree_1): Remove
21019 LTO_STREAMER_DEBUG.
21020 * lto-streamer.c (struct tree_hash_entry): Likewise.
21021 (struct tree_entry_hasher): Likewise.
21022 (tree_entry_hasher::hash): Likewise.
21023 (tree_entry_hasher::equal): Likewise.
21024 (lto_streamer_init): Likewise.
21025 (lto_orig_address_map): Likewise.
21026 (lto_orig_address_get): Likewise.
21027 (lto_orig_address_remove): Likewise.
21028 * lto-streamer.h: Likewise.
21029 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
21030 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
21032 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
21034 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
21035 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
21036 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
21037 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
21038 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
21039 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
21040 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
21041 (mm_maskz_sub_ss): New intrinsics.
21042 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21043 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
21044 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
21045 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
21046 (__builtin_ia32_subss_mask_round): New builtins.
21047 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21048 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
21049 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
21051 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
21052 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
21054 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
21057 2017-05-02 Martin Jambor <mjambor@suse.cz>
21059 PR tree-optimization/78687
21060 * tree-sra.c (access): New field parent.
21061 (process_subtree_disqualification): New function.
21062 (disqualify_candidate): Call it.
21063 (build_accesses_from_assign): Reset write flag if creating an
21065 (build_access_subtree): Fill in parent field and also prpagate
21066 down grp_write flag.
21067 (create_artificial_child_access): New parameter set_grp_write, set
21068 grp_write to its value.
21069 (propagate_subaccesses_across_link): Also propagate grp_write flag
21071 (propagate_all_subaccesses): Push the closest parent back to work
21072 queue if add_access_to_work_queue returned true.
21074 2017-05-02 Richard Biener <rguenther@suse.de>
21076 * common.opt (fstrict-overflow): Alias negative to fwrapv.
21077 * doc/invoke.texi (fstrict-overflow): Remove all traces of
21078 -fstrict-overflow documentation.
21079 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
21080 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
21081 flag_strict_overflow.
21082 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
21083 * lto-opts.c (lto_write_options): Do not stream it.
21084 * lto-wrapper.c (merge_and_complain): Do not handle it.
21085 * opts.c (default_options_table): Do not set -fstrict-overflow.
21086 (finish_options): Likewise do not clear it when sanitizing.
21087 * simplify-rtx.c (simplify_const_relational_operation): Do not
21088 test flag_strict_overflow.
21090 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
21092 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
21093 using enabled attribute.
21094 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
21095 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
21096 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
21097 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
21098 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
21099 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
21100 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
21101 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
21102 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
21103 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
21105 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
21107 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
21109 2017-05-02 Richard Biener <rguenther@suse.de>
21111 PR tree-optimization/80591
21113 2017-04-10 Richard Biener <rguenther@suse.de>
21115 * tree-ssa-structalias.c (find_func_aliases): Properly handle
21118 2017-05-02 Richard Biener <rguenther@suse.de>
21120 PR tree-optimization/80549
21121 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
21122 (cleanup_tree_cfg_noloop): Create forwarders to known loop
21123 headers if they do not have a preheader.
21125 2017-05-02 Martin Liska <mliska@suse.cz>
21128 * common.opt: Fix typo.
21129 * doc/invoke.texi: Likewise.
21131 2017-05-01 Jan Beulich <jbeulich@suse.com>
21133 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
21134 swapping, add (x,x,m,x,n) alternative.
21136 2017-05-01 Nathan Sidwell <nathan@acm.org>
21138 * calls.c (combine_pending_stack_adjustment_and_call): Remove
21139 unnecessary unadjusted_alignment check.
21141 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
21144 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
21146 * gimplify.c (gimplify_cilk_detach): New function.
21147 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
21148 * tree-core.h: Document EXPR_CILK_SPAWN.
21149 * tree.h (EXPR_CILK_SPAWN): Define.
21151 2017-05-01 David Malcolm <dmalcolm@redhat.com>
21153 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
21154 to use new fixit_hint representation, using the "replace" logic.
21155 (get_line_span_for_fixit_hint): Likewise.
21156 (layout::print_any_fixits): Likewise.
21157 (selftest::test_one_liner_many_fixits): Rename to...
21158 (selftest::test_one_liner_many_fixits_1): ...this, and update
21159 comment and expected output to reflect that the multiple fix-it
21160 hints are now consolidated into one insertion.
21161 (selftest::test_one_liner_many_fixits_2): New test.
21162 (selftest::test_diagnostic_show_locus_one_liner): Update for
21164 (selftest::test_fixit_consolidation): Update for fix-it API
21166 * diagnostic.c (print_parseable_fixits): Likewise.
21167 * edit-context.c (edited_line::m_line_events): Convert from
21168 auto_vec <line_event *> to auto_vec <line_event>.
21169 (class line_event): Convert from abstract base class to a concrete
21170 class, taking over the role of replace_event.
21171 (class insert_event): Delete.
21172 (class replace_event): Rename to class line_event. Convert to
21174 (edit_context::add_fixits): Reimplement.
21175 (edit_context::apply_insert): Delete.
21176 (edit_context::apply_replace): Rename to...
21177 (edit_context::apply_fixit): ...this. Convert to half-open range.
21178 (edited_file::apply_insert): Delete.
21179 (edited_file::apply_replace): Rename to...
21180 (edited_file::apply_fixit): ...this.
21181 (edited_line::~edited_line): Drop deletion of events.
21182 (edited_line::apply_insert): Delete.
21183 (edited_line::apply_replace): Rename to...
21184 (edited_line::apply_fixit): ...this. Convert to half-open range.
21185 Update for change to type of m_line_events.
21186 * edit-context.h (edit_context::apply_insert): Delete.
21187 (edit_context::apply_replace): Rename to...
21188 (edit_context::apply_fixit): ...this.
21190 2017-05-01 Martin Sebor <msebor@redhat.com>
21192 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
21195 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
21198 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
21199 __get_cpuid_max returns 0.
21200 (__get_cpuid_count): Ditto.
21202 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
21204 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
21205 replacement expression is another instance of one of its arguments.
21207 2017-05-01 Jakub Jelinek <jakub@redhat.com>
21210 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
21211 check for stack push/pop autoinc.
21212 * config/i386/i386.c (ix86_agi_dependent): Return false
21213 if the only reason why modified_in_p returned true is that
21214 addr is SP based and set_insn is a push or pop.
21216 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
21218 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
21221 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
21224 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
21225 (account_size_time): Use two predicates - exec_pred and
21227 (evaluate_conditions_for_known_args): Compute both clause and
21229 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
21230 (inline_summary_t::duplicate): Update.
21231 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
21233 (compute_inline_parameters): Likewise.
21234 (estimate_edge_size_and_time): Update caluclation of time.
21235 (estimate_node_size_and_time): Compute both time and nonspecialized
21237 (estimate_ipcp_clone_size_and_time): Update.
21238 (inline_merge_summary): Update.
21239 (do_estimate_edge_time): Update.
21240 (do_estimate_edge_size): Update.
21241 (do_estimate_edge_hints): Update.
21242 (inline_read_section, inline_write_summary): Stream both new predicates.
21243 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
21245 (compute_inlined_call_time): Cleanup.
21246 (big_speedup_p): Update.
21247 (edge_badness): Update.
21248 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
21249 (size_time_entry): Replace predicate by exec_predicate and
21250 nonconst_predicate.
21251 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
21252 (estimate_edge_time): Return also nonspec_time.
21253 (reset_edge_growth_cache): Update.
21255 2017-04-29 Jakub Jelinek <jakub@redhat.com>
21257 PR rtl-optimization/80491
21258 * ifcvt.c (noce_process_if_block): When looking for x setter
21259 with missing else_bb, don't check only the insn right before
21260 cond_earliest, but look for the last insn that x is modified in
21261 within the same bb.
21263 PR rtl-optimization/80491
21264 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
21266 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
21268 PR tree-optimization/80487
21269 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
21271 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21273 PR tree-optimization/79697
21274 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
21275 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
21276 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
21278 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
21279 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
21281 2017-04-28 Martin Sebor <msebor@redhat.com>
21283 PR tree-optimization/80523
21284 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
21285 (init_target_to_host_charmap, target_to_host, target_strtol10): New
21287 (maybe_warn, format_directive, parse_directive): Use new functions.
21288 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
21290 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
21292 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
21294 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
21296 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
21297 target_header_dir): Set correctly.
21298 * configure: Regenerated.
21299 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21300 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21301 instead of SYSTEM_HEADER_DIR.
21303 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
21305 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
21306 (estimate_local_effects): Likewise.
21307 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
21308 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
21309 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
21310 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
21311 do_estimate_edge_time, estimate_edge_time): Likewise.
21312 * ipa-inline-analysis.c (estimate_node_size_and_time,
21313 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
21314 (estimate_time_after_inlining): Remove.
21316 2017-04-28 Martin Liska <mliska@suse.cz>
21318 * doc/gcov.texi: Enhance documentation of gcov.
21320 2017-04-28 Martin Liska <mliska@suse.cz>
21322 * doc/gcov.texi: Sort options in alphabetic order.
21323 * doc/gcov-dump.texi: Likewise.
21324 * doc/gcov-tool.texi: Likewise.
21325 * gcov.c (print_usage): Likewise.
21326 * gcov-dump.c (print_usage): Likewise.
21327 * gcov-tool.c (print_merge_usage_message): Likewise.
21328 (print_rewrite_usage_message): Likewise.
21329 (print_overlap_usage_message): Likewise.
21331 2017-04-28 Martin Liska <mliska@suse.cz>
21333 PR gcov-profile/53915
21334 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
21336 2017-04-28 Martin Liska <mliska@suse.cz>
21338 PR gcov-profile/79891
21339 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
21340 is marked by compiler as living on a line.
21341 (get_cycles_count): Remove usage of the union.
21342 (output_intermediate_file): Likewise.
21343 (find_source): Fix GNU coding style.
21344 (accumulate_line_counts): Remove old non-all block mode.
21345 (output_lines): Remove usage of the union.
21346 * profile.c (output_location): Include all BBs, even if
21347 belonging to a same line (and file) as a previous BB.
21349 2017-04-28 Martin Liska <mliska@suse.cz>
21351 * gcov.c (process_args): Handle new argument 'w'.
21352 (read_graph_file): Assign ID to BBs.
21353 (output_branch_count): Display BB # if verbose flag is set.
21354 (output_lines): Likewise for arcs.
21355 (print_usage): Add '--verbose' option help.
21356 * doc/gcov.texi: Document --verbose (-w) option.
21358 2017-04-28 Martin Liska <mliska@suse.cz>
21360 * gcov.c (struct block_location_info): New struct.
21361 (process_file): Fill up the new structure.
21362 (read_graph_file): Replace usage of encoding by the newly added
21364 (add_line_counts): Likewise.
21365 (accumulate_line_counts): Remove usage of the union.
21366 (function_info::function_info): New function.
21367 (function_info::~function_info): Likewise.
21368 (process_file): Call delete instead of release_function.
21369 (release_function): Release the function.
21370 (release_structures): Call delete instead of release_function.
21371 (solve_flow_graph): Replace usage of num_blocks.
21372 (find_exception_blocks): Likewise.
21373 (output_lines): Fix GNU coding style.
21375 2017-04-28 Martin Liska <mliska@suse.cz>
21378 * coverage.c (coverage_remove_note_file): New function.
21379 * coverage.h: Declare the function.
21380 * toplev.c (finalize): Clean if an error has been seen.
21382 2017-04-28 Martin Liska <mliska@suse.cz>
21384 PR gcov-profile/80031
21385 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
21386 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
21387 * gcov.c (read_graph_file): Read just number of blocks.
21388 * profile.c (branch_prob): Do not stream 0 flags per a basic
21391 2017-04-28 Martin Liska <mliska@suse.cz>
21393 * gcov-dump.c (tag_*): Add new argument to declarations.
21394 (dump_gcov_file): Likewise.
21395 (tag_blocks): Add and use new argument depth.
21396 (tag_arcs): Likewise.
21397 (tag_lines): Likewise.
21398 (tag_counters): Likewise.
21399 (tag_summary): Likewise.
21400 (dump_working_sets): Use depth to do a proper indentation.
21402 2017-04-28 Jakub Jelinek <jakub@redhat.com>
21405 * cgraph.h (symtab_node::debug_symtab): No longer inline.
21406 * symtab.c (symtab_node::debug_symtab): Move definition here.
21408 2017-04-28 Richard Biener <rguenther@suse.de>
21410 * lto-streamer.h (LTO_major_version): Bump to 7.
21412 2017-04-28 Richard Biener <rguenther@suse.de>
21414 * tree-vrp.c (assert_info): New struct.
21415 (add_assert_info): New helper.
21416 (register_edge_assert_for_2): Refactor to add asserts to a vector
21418 (register_edge_assert_for_1): Likewise.
21419 (register_edge_assert_for): Likewise.
21420 (finish_register_edge_assert_for): New helper actually registering
21421 asserts where live on edge.
21422 (find_conditional_asserts): Adjust.
21423 (find_switch_asserts): Likewise.
21424 (evrp_dom_walker::try_find_new_range): Generalize.
21425 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
21427 2017-04-27 Marek Polacek <polacek@redhat.com>
21430 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
21431 arg10 and arg11 to itype.
21433 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
21435 * doc/extend.texi (Object Size Checking): Improve grammar.
21437 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
21440 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
21441 that the logic for permitting reciprocal estimates matches that
21444 2017-04-27 Jakub Jelinek <jakub@redhat.com>
21447 * tree.c (type_cache_hasher::equal): Only compare
21448 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
21449 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
21450 non-aggregate element types.
21451 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
21452 about the flag on ARRAY_TYPEs in the comment, formatting fix.
21454 2017-04-27 Richard Biener <rguenther@suse.de>
21456 PR middle-end/80533
21457 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
21458 stripping ARRAY_REFs from MEM_EXPR make sure we're not
21459 keeping a reference to a trailing array.
21461 2017-04-27 Richard Biener <rguenther@suse.de>
21463 PR middle-end/80539
21464 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
21465 being in loop-closed SSA form conservatively.
21466 (chrec_fold_multiply_poly_poly): Likewise.
21468 2017-04-27 Tamar Christina <tamar.christina@arm.com>
21470 PR middle-end/79665
21471 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
21472 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
21474 2017-04-27 Jakub Jelinek <jakub@redhat.com>
21477 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
21478 (aarch64_function_arg_alignment): Return unsigned int again, but still
21479 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
21480 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
21481 Don't emit -Wpsabi note.
21482 (aarch64_function_arg_boundary): Likewise.
21483 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
21486 2017-04-26 Nathan Sidwell <nathan@acm.org>
21488 * tree.h (crc32_unsigned_n): Declare.
21489 (crc32_unsigned, crc32_unsigned): Make inline.
21490 * tree.c (crc32_unsigned_bits): Replace with ...
21491 (crc32_unsigned_n): ... this.
21492 (crc32_unsigned, crc32_byte): Remove.
21493 (crc32_string): Remove unnecessary braces.
21495 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
21497 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
21498 * ipa-inline-analysis.c (MAX_TIME): Remove.
21499 (account_size_time): Use sreal for time.
21500 (dump_inline_summary): Update.
21501 (estimate_function_body_sizes): Update.
21502 (estimate_edge_size_and_time): Update.
21503 (estimate_calls_size_and_time): Update.
21504 (estimate_node_size_and_time): Update.
21505 (inline_merge_summary): Update.
21506 (inline_update_overall_summary): Update.
21507 (estimate_time_after_inlining): Update.
21508 (inline_read_section): Update.
21509 (inline_write_summary): Update.
21510 * ipa-inline.c (compute_uninlined_call_time): Update.
21511 (compute_inlined_call_time): Update.
21512 (recursive_inlining): Update.
21513 (inline_small_functions): Update.
21514 (dump_overall_stats): Update.
21515 * ipa-inline.h: Include sreal.h.
21516 (size_time_entry): Turn time to sreal.
21517 (inline_summary): Turn self_time nad time to sreal.
21519 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
21521 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
21523 (sreal::stream_out, sreal::stream_in): New.
21524 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
21526 2017-04-25 Jakub Jelinek <jakub@redhat.com>
21528 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
21531 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
21534 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
21536 (dimode_scalar_chain::compute_convert_gain): Ditto.
21537 (dimode_scalar_chain::make_vector_copies): Ditto.
21538 (dimode_scalar_chain::convert_reg): Ditto.
21539 (dimode_scalar_chain::convert_insn): Ditto.
21540 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
21541 (VI248_AVX512BW_1): New mode iterator.
21542 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
21543 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
21546 2017-04-25 Martin Sebor <msebor@redhat.com>
21548 PR tree-optimization/80497
21549 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
21550 constants are representable in HOST_WIDE_INT.
21551 (parse_directive): Ditto.
21553 2017-04-25 Martin Sebor <msebor@redhat.com>
21556 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
21557 (new_zero_array): Adjust signature.
21558 (dom_info::dom_init): Used unsigned rather that size_t.
21559 (dom_info::dom_info): Same.
21561 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21562 Jakub Jelinek <jakub@redhat.com>
21565 * config/arm/arm.c: Include gimple.h.
21566 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
21567 returns negative, increment ncrn only if it returned positive.
21568 (arm_needs_doubleword_align): Return int instead of bool,
21569 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
21570 members, but if there is any such non-FIELD_DECL
21571 > PARM_BOUNDARY aligned decl, return -1 instead of false.
21572 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
21573 returns negative, increment nregs only if it returned positive.
21574 (arm_setup_incoming_varargs): Likewise.
21575 (arm_function_arg_boundary): Emit -Wpsabi note if
21576 arm_needs_doubleword_align returns negative, return
21577 DOUBLEWORD_ALIGNMENT only if it returned positive.
21579 2017-04-25 Marek Polacek <polacek@redhat.com>
21582 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
21583 first argument to type.
21585 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
21588 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
21589 type checks to test for compatibility instead of equality.
21591 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21592 Jakub Jelinek <jakub@redhat.com>
21595 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
21597 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
21598 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
21599 the alignment computation, but return their maximum in warn_alignment.
21600 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
21601 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
21603 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
21604 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
21607 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21609 * config/arc/simdext.md (dmpyh): Fix typo.
21611 2017-04-25 Richard Biener <rguenther@suse.de>
21613 PR tree-optimization/80492
21614 * alias.c (compare_base_decls): Handle registers with asm
21615 specification conservatively.
21616 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
21617 compare_base_decls returning dont-know properly.
21619 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21621 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
21622 (legitimate_offset_address_p): New function.
21623 (arc_legitimate_address_p): Use above function.
21625 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21627 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
21629 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21631 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
21632 ACCH registers whenever they are available.
21634 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21636 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
21637 double regs fix when not used.
21639 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21641 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
21643 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
21644 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
21646 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21648 * config/arc/arc.c (arc_output_addsi): Check for h-register class
21649 when emitting short ADD instructions.
21651 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21653 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
21655 (cmpsi_cc_c_insn): Likewise.
21656 (cbranchsi4_scratch): Compute proper instruction length using
21657 compact_hreg_operand.
21658 * config/arc/predicates.md (compact_hreg_operand): New predicate.
21660 2017-04-25 Richard Biener <rguenther@suse.de>
21662 PR middle-end/80509
21663 * passes.c (pass_manager::pass_manager): Initialize
21664 m_name_to_pass_map.
21666 2017-04-25 Richard Biener <rguenther@suse.de>
21668 PR tree-optimization/79201
21669 * tree-ssa-sink.c (statement_sink_location): Handle calls.
21671 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21674 * config/s390/vector.md: Split MEM->GPR vector moves for
21675 non-s_operand addresses.
21677 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21680 * config/s390/predicates.md (reload_const_wide_int_operand): New
21682 * config/s390/s390.md ("movti"): Remove d/P alternative.
21683 ("movti_bigconst"): New pattern definition.
21685 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
21688 * s390-protos.h (s390_expand_cs_hqi): Removed.
21689 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
21690 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
21691 modes as well as CCZ1mode and CCZmode.
21692 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
21693 signature of s390_emit_compare_and_swap.
21694 (s390_expand_cs_hqi): Likewise, make static.
21695 (s390_expand_cs_tdsi): Generate an explicit compare before trying
21696 compare-and-swap, in some cases.
21697 (s390_expand_cs): Wrapper function.
21698 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
21700 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
21701 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
21702 patterns for small and large integers. Forbid symref memory operands.
21703 Move expander to s390.c. Require cc register.
21704 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
21705 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
21706 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
21707 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
21708 symref memory operands. Remove CC mode and call s390_match_ccmode
21710 ("atomic_exchange<mode>"): Allow and implement all integer modes.
21712 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
21714 * config/s390/s390.md (define_peephole2): New peephole to help
21715 combining the load-and-test pattern with volatile memory.
21717 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
21719 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
21720 with CCZmode for TARGET_Z196.
21722 2017-04-25 Jakub Jelinek <jakub@redhat.com>
21724 PR rtl-optimization/80501
21725 * combine.c (make_compound_operation_int): Set subreg_code to SET
21726 even for AND with mask of the sign bit of mode.
21728 PR rtl-optimization/80500
21729 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
21730 sum's initial value.
21732 2017-04-25 Julian Brown <julian@codesourcery.com>
21733 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21735 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
21737 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
21739 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
21741 2017-04-25 Julian Brown <julian@codesourcery.com>
21742 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21744 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
21745 (thunderx2t99_sha): New Reservation.
21747 2017-04-25 Julian Brown <julian@codesourcery.com>
21748 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21750 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
21751 type for 1-element load.
21753 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
21755 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
21757 2017-04-24 Martin Jambor <mjambor@suse.cz>
21759 PR tree-optimization/80293
21760 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
21761 char arrays not totally scalarizable if it is false.
21762 (analyze_all_variable_accesses): Pass correct value in the new
21763 parameter. Add a statistics counter.
21765 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
21767 PR middle-end/79931
21768 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
21770 2017-04-24 Richard Biener <rguenther@suse.de>
21772 PR tree-optimization/80494
21773 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
21774 out for complex types.
21776 2017-04-24 Richard Biener <rguenther@suse.de>
21778 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
21779 * tree-ssa-sccvn.c (print_scc): Print SCC size.
21780 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
21781 (DFS): Adjust and never fail.
21782 (sccvn_dom_walker::fail): Remove.
21783 (sccvn_dom_walker::before_dom_children): Adjust.
21784 (run_scc_vn): Likewise and never fail.
21785 * tree-ssa-pre.c (pass_pre::execute): Adjust.
21786 (pass_fre::execute): Likewise.
21788 2017-04-24 Richard Biener <rguenther@suse.de>
21790 PR tree-optimization/79725
21791 * tree-ssa-sink.c (statement_sink_location): Return whether
21792 failure reason was zero uses. Move that check later.
21793 (sink_code_in_bb): Deal with zero uses by removing the stmt
21796 2017-04-24 Richard Biener <rguenther@suse.de>
21799 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
21800 pointer-based references.
21802 2017-04-24 Richard Biener <rguenther@suse.de>
21805 * pass_manager.h (pass_manager::operator new): Remove.
21806 (pass_manager::operator delete): Likewise.
21807 * passes.c (pass_manager::operator new): Remove.
21808 (pass_manager::operator delete): Likewise.
21809 (pass_manager::pass_manager): Zero individual pass members.
21811 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
21814 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
21815 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
21816 Check "XEXP (src, 1)" operand here.
21817 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
21818 Check "XEXP (src, 1)" operand here.
21819 (dimode_scalar_chain::make_vector_copies): Detect count register
21820 of a shift instruction. Zero extend count register from QImode
21821 to DImode to satisfy vector shift pattern count operand predicate.
21822 Substitute vector shift count operand with a DImode copy.
21823 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
21826 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
21828 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
21829 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
21830 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
21831 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
21832 (UNSPEC_NOREX_MEM): Remove definition.
21834 2017-04-21 Richard Biener <rguenther@suse.de>
21836 PR tree-optimization/79547
21837 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
21838 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
21839 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
21840 without any constraints.
21842 2017-04-21 Richard Biener <rguenther@suse.de>
21844 PR tree-optimization/78847
21845 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
21847 2017-04-21 Richard Biener <rguenther@suse.de>
21849 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
21850 (build_distinct_type_copy): Likewise.
21851 (build_variant_type_copy): Likewise.
21852 * tree.c (build_qualified_type): Pass down mem-stat info.
21853 (build_distinct_type_copy): Likewise.
21854 (build_variant_type_copy): Likewise.
21856 2017-04-21 Richard Biener <rguenther@suse.de>
21858 PR tree-optimization/80237
21859 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
21861 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
21862 for a simplified result.
21864 2016-04-21 Richard Biener <rguenther@suse.de>
21866 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
21867 sth as strict as a simple_iv but a chrec without symbols and an
21868 operand defined in the loop we are peeling (and not some subloop).
21869 (propagate_constants_for_unrolling): Propagate all constants.
21871 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
21874 * config/i386/i386.c (print_reg): Remove assert for disalowed
21875 regno values, call output_operand_lossage instead.
21877 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
21880 * config/i386/constraints.md (Yc): New register constraint.
21881 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
21882 Use Yc constraint for alternative 2 of operand 0. Remove
21883 preferred_for_speed attribute.
21885 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
21887 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
21888 lastprivate clauses in SIMT case.
21890 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
21892 * doc/invoke.texi (-Wextra-semi): Document new warning option.
21894 2017-04-20 Richard Biener <rguenther@suse.de>
21896 PR tree-optimization/57796
21897 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
21898 as N scalar stores.
21899 (vect_model_load_cost): Cost gathers as N scalar loads.
21901 2017-04-20 Richard Biener <rguenther@suse.de>
21903 * ggc-page.c (ggc_allocated_p): Rename to ...
21904 (safe_lookup_page_table_entry): ... this and return the lookup
21906 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
21908 2017-04-20 Richard Biener <rguenther@suse.de>
21910 PR tree-optimization/80453
21911 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
21912 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
21913 from the conditions.
21914 (vn_phi_eq): Pass them down.
21915 (vn_phi_lookup): Record them.
21916 (vn_phi_insert): Likewise.
21918 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
21920 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
21921 uninitialized variable warning to avoid buffer overrun.
21923 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
21926 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
21927 is suppressed for '{ 0 }' in C.
21929 2017-04-20 Jakub Jelinek <jakub@redhat.com>
21931 * BASE-VER: Set to 8.0.0.
21933 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
21935 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
21936 priority .init_array and .fini_array section with SECTION_NOTYPE
21939 2017-04-20 Jakub Jelinek <jakub@redhat.com>
21941 PR middle-end/80423
21942 * tree.h (build_array_type): Add typeless_storage default argument.
21943 * tree.c (type_cache_hasher::equal): Also compare
21944 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
21945 (build_array_type): Add typeless_storage argument, set
21946 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
21948 (build_nonshared_array_type): Adjust build_array_type_1 caller.
21949 (build_array_type): Likewise. Add typeless_storage argument.
21951 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
21952 Jakub Jelinek <jakub@redhat.com>
21954 PR tree-optimization/80426
21955 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
21956 operation on symbolic operands, also compute the overflow for the
21957 invariant part when the operation degenerates into a negation.
21959 2017-04-19 Jakub Jelinek <jakub@redhat.com>
21962 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
21963 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
21966 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
21968 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
21971 * config/avr/avr.c (tree.h): Include it.
21972 (cgraph.h): Include it.
21973 (avr_encode_section_info): Don't warn for uninitialized progmem
21974 variable if it's just an alias.
21976 2017-04-19 Richard Biener <rguenther@suse.de>
21979 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
21980 when needed by AutoPGO.
21982 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
21985 * doc/lto.texi: Remove an extra 'that'.
21987 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
21989 PR rtl-optimization/80429
21990 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
21991 are only used in debug insns.
21993 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
21994 Vladimir Makarov <vmakarov@redhat.com>
21996 * config/sparc/predicates.md (input_operand): Add comment. Return
21997 true for any memory operand when LRA is in progress.
21998 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
22000 2017-04-18 Jeff Law <law@redhat.com>
22003 * mips.md ({return,simple_return}_internal): Do not overwrite
22006 2017-04-18 Jakub Jelinek <jakub@redhat.com>
22008 PR tree-optimization/80443
22009 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
22010 instead of adding 1, subtract -1 and similarly instead of subtracting
22013 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
22015 PR rtl-optimization/80357
22016 * haifa-sched.c (tmp_bitmap): New variable.
22017 (model_recompute): Handle duplicate use records.
22018 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
22019 (free_global_sched_pressure_data): Free it.
22021 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
22024 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
22025 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
22026 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
22027 instead of SYSTEM_HEADER_DIR.
22029 2017-04-18 Jeff Law <law@redhat.com>
22031 PR middle-end/80422
22032 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
22033 predecessors after walking up the insn chain.
22035 2017-04-18 Jakub Jelinek <jakub@redhat.com>
22038 * dwarf2out.c (modified_type_die): Try harder not to emit internal
22039 sizetype type into debug info.
22041 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22044 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
22045 unneeded test for TARGET_UPPER_REGS_SF.
22046 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
22048 2017-04-18 Jakub Jelinek <jakub@redhat.com>
22051 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
22052 instead of gsi_after_labels.
22054 2017-04-18 Jeff Law <law@redhat.com>
22056 * regcprop.c (maybe_mode_change): Avoid creating copies of the
22060 2017-04-13 Jeff Law <law@redhat.com>
22061 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
22062 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
22064 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
22067 * config/avr/avr.c (intl.h): Include it.
22068 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
22070 2017-04-18 Martin Liska <mliska@suse.cz>
22072 PR gcov-profile/78783
22073 * gcov-tool.c (gcov_output_files): Validate that destination
22074 file is either removed by the tool or by a user.
22076 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
22077 Guy Benyei <guybe@mellanox.com>
22079 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
22080 block, and do not negate it, the stored id is already negative.
22082 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
22084 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
22086 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
22089 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
22090 masks of options that should be turned off if the VSX vector
22091 options are turned off.
22092 (OTHER_P8_VECTOR_MASKS): Likewise.
22093 (OTHER_VSX_VECTOR_MASKS): Likewise.
22094 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
22095 rs6000_disable_incompatible_switches to validate no type switches
22097 (rs6000_incompatible_switch): New function to disallow turning on
22098 other vector options if -mno-vsx, -mno-power8-vector, or
22099 -mno-power9-vector are specified.
22101 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
22103 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
22105 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
22107 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
22108 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
22109 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
22110 (ARG_POINTER_CFA_OFFSET): Likewise.
22112 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
22114 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
22115 conditions to take advantage of various optimizations.
22117 2017-04-13 Jeff Law <law@redhat.com>
22119 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
22120 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
22121 (zero_extendsidi2_dext): Likewise.
22123 2017-04-13 Jakub Jelinek <jakub@redhat.com>
22126 * fold-const.c (fold_ternary_loc): Revert
22127 use op0 instead of fold_convert_loc (loc, type, arg0) part of
22130 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
22132 PR rtl-optimization/80343
22133 * lra-remat.c (update_scratch_ops): Assign original hard reg to
22134 new scratch pseudo.
22136 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
22139 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
22140 to ubsan_encode_value.
22142 2017-04-13 Jeff Law <law@redhat.com>
22144 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
22145 appearing in DEBUG_INSNs.
22147 2017-04-13 Martin Liska <mliska@suse.cz>
22149 PR gcov-profile/80413
22150 * gcov-io.c (gcov_write_string): Copy to buffer just when
22151 allocated size is greater than zero.
22153 2017-04-13 Jakub Jelinek <jakub@redhat.com>
22156 * dwarf2out.c (decls_for_scope): Ignore declarations of
22157 current_function_decl in BLOCK_NONLOCALIZED_VARS.
22159 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
22162 * ipa-visibility.c (non_local_p): Fix typos.
22163 (localize_node): When localizing symbol in same comdat group,
22164 dissolve the group only when we know external symbols are going
22166 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
22168 2017-04-12 Jakub Jelinek <jakub@redhat.com>
22170 PR tree-optimization/79390
22171 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
22172 order does not result in usable sequence, retry with reversed operand
22178 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
22179 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
22180 op0 instead of fold_convert_loc (loc, type, arg0).
22182 2017-04-12 Jeff Law <law@redhat.com>
22184 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
22185 has a delay slot in the generated code.
22187 * config/cris/cris.md (cris_preferred_reload_class): Return
22188 GENNONACR_REGS rather than GENERAL_REGS.
22190 2017-04-12 Jakub Jelinek <jakub@redhat.com>
22193 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
22194 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
22195 signedness of the result type.
22197 2017-04-12 Richard Biener <rguenther@suse.de>
22198 Jeff Law <law@redhat.com>
22200 PR tree-optimization/80359
22201 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
22202 trim stores to TARGET_MEM_REFs.
22204 2017-04-12 Richard Biener <rguenther@suse.de>
22206 PR tree-optimization/79390
22207 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
22208 threading case even more.
22210 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
22213 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
22214 for quad_address_p for TImode, instead of just not indexed_address.
22216 2017-04-12 Richard Biener <rguenther@suse.de>
22217 Bernd Edlinger <bernd.edlinger@hotmail.de>
22219 PR middle-end/79671
22220 * alias.c (component_uses_parent_alias_set_from): Handle
22221 TYPE_TYPELESS_STORAGE.
22222 (get_alias_set): Likewise.
22223 * tree-core.h (tree_type_common): Add typeless_storage flag.
22224 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
22225 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
22226 for types containing members with TYPE_TYPELESS_STORAGE.
22227 (place_field): Likewise.
22228 (layout_type): Likewise for ARRAY_TYPE.
22229 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
22230 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
22231 TYPE_TYPELESS_STORAGE.
22232 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
22234 2017-04-12 Jakub Jelinek <jakub@redhat.com>
22237 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
22238 first argument to type.
22240 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22244 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
22245 CONST0_RTX (mode) rather than const0_rtx where appropriate.
22246 (rs6000_expand_binop_builtin): Likewise.
22247 (rs6000_expand_ternop_builtin): Likewise; also add missing
22248 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
22249 vshasigma built-ins.
22250 * doc/extend.texi: Document that vec_xxpermdi's third argument
22251 must be a constant.
22253 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
22255 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
22256 Use shift_const cost parameter when calculating gain of STV shifts.
22258 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
22260 PR rtl-optimization/70478
22261 * lra-constraints.c (process_alt_operands): Check memory for
22262 disfavoring memory insn operand.
22264 2017-04-11 Jakub Jelinek <jakub@redhat.com>
22266 PR middle-end/80100
22267 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
22268 left shift in unsigned HOST_WIDE_INT type.
22270 PR rtl-optimization/80385
22271 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
22272 (not (neg X)) into (plus X -1) for complex or non-integral modes.
22275 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
22276 if they have any depend clauses.
22278 2017-04-11 Martin Liska <mliska@suse.cz>
22281 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
22282 * ipa-split.c (split_function): Create a local comdat symbol
22283 if caller is in a comdat group.
22285 2017-04-11 Martin Liska <mliska@suse.cz>
22288 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
22291 2017-04-11 Martin Sebor <msebor@redhat.com>
22293 PR middle-end/80364
22294 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
22295 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
22297 (directive::set_width, directive::set_precision, format_character):
22299 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
22302 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
22305 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
22306 conflict, set target->arch_name instead of target->cpu_name.
22308 2017-04-11 Richard Biener <rguenther@suse.de>
22310 PR tree-optimization/80374
22311 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
22312 build_zero_cst, remove fold_convertible_p check again.
22314 2017-04-11 Martin Liska <mliska@suse.cz>
22317 * ubsan.c (instrument_object_size): Do not instrument register
22320 2017-04-11 Jakub Jelinek <jakub@redhat.com>
22323 * config/i386/i386-builtin-types.def
22324 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
22325 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
22326 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
22327 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
22328 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
22329 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
22330 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
22331 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
22332 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
22333 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
22334 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
22335 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
22336 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
22337 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
22338 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
22339 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
22340 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
22341 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
22342 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
22343 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
22344 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
22345 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
22346 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
22347 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
22348 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
22349 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
22350 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
22351 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
22352 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
22353 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
22354 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
22355 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
22356 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
22357 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
22358 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
22359 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
22360 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
22361 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
22362 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
22363 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
22364 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
22365 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
22366 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
22367 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
22368 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
22369 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
22371 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
22372 flag to second_arg_count, handle 4 argument function type _COUNT
22373 aliases, handle second_arg_count on second argument rather than last.
22375 2017-04-10 Jeff Law <law@redhat.com>
22377 PR tree-optimization/80374
22378 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
22379 record anything if we can not convert integer_zero_node to the
22382 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
22385 * config/rs6000/rs6000.c (rs6000_option_override_internal):
22386 Enhance special handling given to the TARGET_P9_MINMAX option in
22387 relation to certain other options.
22389 2017-04-10 Bin Cheng <bin.cheng@arm.com>
22391 PR tree-optimization/80153
22392 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
22393 remove POINTER_PLUS_EXPR's base part directly, rather than through
22396 2017-04-10 Richard Biener <rguenther@suse.de>
22397 Bin Cheng <bin.cheng@arm.com>
22399 PR tree-optimization/80153
22400 * tree-affine.c (aff_combination_to_tree): Get base pointer from
22401 the first element of pointer type aff_tree. Build result expr in
22403 (add_elt_to_tree): Convert to type unconditionally. Remove other
22404 fold_convert calls.
22405 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
22406 (rewrite_use_nonlinear_expr): Check invariant using iv information.
22408 2017-04-10 Richard Biener <rguenther@suse.de>
22410 * tree-ssa-structalias.c (find_func_aliases): Properly handle
22413 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
22415 PR rtl-optimization/70478
22416 * lra-constraints.c (curr_small_class_check): New.
22417 (update_and_check_small_class_inputs): New.
22418 (process_alt_operands): Update curr_small_class_check. Disfavor
22419 alternative insn memory operands. Check available regs for small
22422 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
22425 * config/mips/mips.opt (-mvirt): Update description.
22426 * doc/invoke.texi (-mvirt): Likewise.
22428 2017-04-10 Richard Biener <rguenther@suse.de>
22430 PR middle-end/80362
22431 * fold-const.c (fold_binary_loc): Look at unstripped ops when
22432 looking for NEGATE_EXPR in -A / -B to A / B folding.
22434 2017-04-10 Martin Liska <mliska@suse.cz>
22436 PR gcov-profile/80224
22437 * gcov.c (print_usage): Fix usage string.
22438 (get_gcov_intermediate_filename): Remove.
22439 (output_gcov_file): Use both for normal and intermediate format.
22440 (generate_results): Do not initialize special file for
22441 intermediate format.
22443 2017-04-10 Richard Biener <rguenther@suse.de>
22445 PR tree-optimization/80304
22446 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
22449 2017-04-10 Nathan Sidwell <nathan@acm.org>
22452 * config/rs6000/rs6000.c (rs6000_vector_type): New.
22453 (rs6000_init_builtins): Use it.
22455 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22457 * config/arm/arm.md (<mrc>): Add mode to SET source.
22458 (<mrrc>): Likewise.
22460 2017-04-10 Richard Biener <rguenther@suse.de>
22462 PR middle-end/80344
22463 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
22465 2017-04-10 Jakub Jelinek <jakub@redhat.com>
22468 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
22469 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
22470 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
22471 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
22472 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
22473 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
22474 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
22475 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
22476 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
22477 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
22478 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
22479 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
22480 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
22481 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
22482 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
22483 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
22484 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
22485 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
22486 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
22487 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
22488 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
22489 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
22490 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
22492 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
22494 PR rtl-optimization/70478
22495 * lra-constraints.c: Reverse the last patch.
22497 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
22499 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
22500 Add comment for WCHAR_T.
22502 2017-04-08 Martin Liska <mliska@suse.cz>
22505 2017-04-07 Martin Liska <mliska@suse.cz>
22508 * ipa-split.c (split_function): Add function part to a same comdat
22511 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22514 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
22516 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
22518 * rs6000/rs6000.c (vec_load_pendulum): Rename...
22519 (vec_pairing): ...to this.
22520 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
22521 (rs6000_sched_init): Adjust for name change.
22522 (struct rs6000_sched_context): Likewise.
22523 (rs6000_init_sched_context): Likewise.
22524 (rs6000_set_sched_context): Likewise.
22526 2017-04-07 Jakub Jelinek <jakub@redhat.com>
22532 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
22534 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
22535 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
22536 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
22538 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
22540 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
22542 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
22544 PR rtl-optimization/70703
22545 * ira-color.c (update_conflict_hard_regno_costs): Use
22546 int64_t instead of HOST_WIDE_INT.
22548 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
22550 PR rtl-optimization/70478
22551 * lra-constraints.c (process_alt_operands): Disfavor alternative
22552 insn memory operands.
22554 2017-04-07 Jeff Law <law@redhat.com>
22556 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
22557 CALL and NOTE_INSN_CALL_ARG_LOCATION.
22559 2017-04-07 Martin Liska <mliska@suse.cz>
22562 * config/aarch64/aarch64.c (aarch64_process_target_attr):
22563 Show error message instead of an ICE.
22565 2017-04-07 Martin Liska <mliska@suse.cz>
22568 * ipa-split.c (split_function): Add function part to a same comdat
22571 2017-04-07 Richard Biener <rguenther@suse.de>
22573 PR middle-end/80341
22574 * tree.c (get_unwidened): Also handle ! for_type case for
22576 * convert.c (do_narrow): Split out from ...
22577 (convert_to_integer_1): ... here. Do not pass final truncation
22578 type to get_unwidened for TRUNC_DIV_EXPR.
22580 2017-04-07 Richard Biener <rguenther@suse.de>
22582 * tree-affine.c (wide_int_ext_for_comb): Take type rather
22584 (aff_combination_const): Adjust.
22585 (aff_combination_scale): Likewise.
22586 (aff_combination_add_elt): Likewise.
22587 (aff_combination_add_cst): Likewise.
22588 (aff_combination_convert): Likewise.
22589 (add_elt_to_tree): Likewise. Remove unused argument.
22590 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
22592 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
22594 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
22596 * config/arm/arm.c (arm_default_short_enums): Use
22597 ARM_DEFAULT_SHORT_ENUMS.
22598 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
22600 2017-04-06 Jakub Jelinek <jakub@redhat.com>
22603 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
22604 members with redundant out-of-class redeclaration.
22606 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
22609 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
22610 * config/i386/i386.md (*zero_extendsidi2):
22611 Add (?*x,*x) and (?*v,*v) alternatives.
22613 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
22616 * config/i386/i386.c (ix86_expand_builtin)
22617 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
22618 mode from insn data. Convert operands to insn operand mode.
22619 Copy operands that don't satisfy insn predicate to a register.
22621 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
22623 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
22626 2017-04-06 Richard Biener <rguenther@suse.de>
22628 PR tree-optimization/80334
22629 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
22630 preserve alignment of accesses.
22632 2017-04-06 Richard Biener <rguenther@suse.de>
22634 PR tree-optimization/80262
22635 * tree-sra.c (build_ref_for_offset): Preserve address-space
22637 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
22638 Drop useless address-space information on MEM_REF offsets.
22640 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
22642 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
22644 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
22646 PR rtl-optimization/70703
22647 * ira-color.c (update_conflict_hard_regno_costs): Use
22648 HOST_WIDE_INT instead of long.
22650 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
22653 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
22654 not defined for x86_64 target. Add -mmmx target option when __SSE2__
22656 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
22657 for x86_64 target. Handle -m3dnowa option.
22659 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
22661 PR rtl-optimization/70703
22662 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
22663 (update_conflict_hard_regno_costs): Use long instead of unsigned
22664 arithmetic for cost calculation.
22666 2017-04-05 Jakub Jelinek <jakub@redhat.com>
22667 Bernd Edlinger <bernd.edlinger@hotmail.de>
22670 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
22673 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
22676 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
22677 ptr_mode with Pmode throughout.
22678 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
22679 into probe_stack_range and use DImode.
22681 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
22684 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
22685 call_eh_return is true.
22687 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22689 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
22690 Initialize last_match_fntype_index.
22692 2017-04-05 Jakub Jelinek <jakub@redhat.com>
22695 * tree-nvr.c: Include internal-fn.h.
22696 (pass_return_slot::execute): Ignore internal calls without
22699 2017-04-04 Jakub Jelinek <jakub@redhat.com>
22700 Richard Biener <rguenther@suse.de>
22703 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
22704 captures used multiple times, except for the last use.
22705 * generic-match-head.c: Include gimplify.h.
22707 2017-04-04 Jakub Jelinek <jakub@redhat.com>
22709 PR tree-optimization/79390
22710 * target.h (struct noce_if_info): Declare.
22711 * targhooks.h (default_noce_conversion_profitable_p): Declare.
22712 * target.def (noce_conversion_profitable_p): New target hook.
22713 * ifcvt.h (struct noce_if_info): New type, moved from ...
22714 * ifcvt.c (struct noce_if_info): ... here.
22715 (noce_conversion_profitable_p): Renamed to ...
22716 (default_noce_conversion_profitable_p): ... this. No longer
22718 (noce_try_store_flag_constants, noce_try_addcc,
22719 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
22720 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
22721 instead of noce_conversion_profitable_p.
22722 * config/i386/i386.c: Include ifcvt.h.
22723 (ix86_option_override_internal): Don't override
22724 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
22725 (ix86_noce_conversion_profitable_p): New function.
22726 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
22727 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
22728 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
22729 * doc/tm.texi: Regenerated.
22731 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22733 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
22736 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22739 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
22740 instructions for small multiply cores.
22742 2017-04-04 Jeff Law <law@redhat.com>
22744 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
22746 (mips_expand_vec_perm_const): Initialize elements in orig_perm
22747 that are not set by the loop over the elements.
22749 2017-04-04 Jakub Jelinek <jakub@redhat.com>
22752 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
22753 int mode, convert_modes it to mode as unsigned, otherwise use
22754 lowpart_subreg to mode rather than SImode.
22755 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
22756 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
22757 Use DImode instead of SImode for the shift count operand.
22758 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
22761 2017-04-04 Richard Biener <rguenther@suse.de>
22763 PR middle-end/80281
22764 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
22765 arithmetic done for the negate or the plus. Simplify.
22766 (A - (-B) -> A + B): Likewise.
22767 * fold-const.c (split_tree): Make sure to not negate pointers.
22769 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
22771 PR rtl-optimization/60818
22772 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
22773 a compare of comparisons with the thing compared if this results
22774 in a different machine mode.
22776 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
22778 * alias.c (base_alias_check): Fix typo in comment.
22779 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
22780 * cgraphunit.c (symbol_table::compile): Likewise.
22781 * collect2.c (maybe_run_lto_and_relink): Likewise.
22782 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
22783 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
22784 * config/avr/avr.c (avr_map_op_t): Likewise.
22785 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
22786 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
22787 * config/epiphany/epiphany.md (movcc): Likewise.
22788 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
22789 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
22791 * config/mips/mips.c (mips_save_restore_reg): Likewise.
22792 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
22793 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
22794 * config/sh/sh.c (sh_rtx_costs): Likewise.
22795 * fold-const.c (fold_truth_andor): Likewise.
22796 * genautomata.c (collapse_flag): Likewise.
22797 * gengtype.h (struct type::u::s): Likewise.
22798 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
22799 * input.c (FORMAT_AMOUNT): Likewise.
22800 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
22801 (known_aggs_to_agg_replacement_list): Likewise.
22802 * ipa-inline-analysis.c: Likewise.
22803 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
22804 * ipa-polymorphic-call.c
22805 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
22806 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
22807 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
22809 * modulo-sched.c (apply_reg_moves): Likewise.
22810 * omp-expand.c (build_omp_regions_1): Likewise.
22811 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
22812 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
22813 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
22814 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
22815 * value-prof.c: Likewise.
22816 * var-tracking.c (val_reset): Likewise.
22818 2017-04-03 Richard Biener <rguenther@suse.de>
22820 PR tree-optimization/80275
22821 * fold-const.c (split_address_to_core_and_offset): Handle
22824 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
22826 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
22827 descriptors is at least equal to that of functions.
22829 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
22831 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
22833 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
22836 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
22837 (mov<IMOD4:mode>): New expander.
22838 (*mov<IMOD4:mode>_internal): New insn and split pattern.
22840 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
22842 PR rtl-optimization/79405
22843 * fwprop.c (propagations_left): New variable.
22844 (forward_propagate_into): Decrement it.
22845 (fwprop_init): Initialize it.
22846 (fw_prop): If the variable has reached zero, stop propagating.
22847 (fwprop_addr): Ditto.
22849 2017-03-31 Jakub Jelinek <jakub@redhat.com>
22852 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
22853 a FUNCTION_DECL, pass it as decl instead of origin to
22856 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
22858 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
22861 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
22864 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
22865 TARGET_VSX_SMALL_INTEGER.
22867 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22869 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
22870 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
22872 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
22874 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
22875 extraction from odd-numbered MSA register.
22877 2017-03-31 Jakub Jelinek <jakub@redhat.com>
22879 PR middle-end/80173
22880 * expmed.c (store_bit_field_1): Don't attempt to create
22881 a word subreg out of hard registers wider than word if they
22882 have HARD_REGNO_NREGS of 1 for their mode.
22884 PR middle-end/80163
22885 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
22886 conversions to integer types wider than word and pointer.
22889 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
22890 (rtx_equal_for_cselib_p): Pass 0 to it.
22891 * cselib.c (cselib_hasher::equal): Likewise.
22892 (rtx_equal_for_cselib_1): Add depth argument. If depth
22893 is 128, don't look up VALUE locs and punt. Increment
22894 depth in recursive calls when walking VALUE locs.
22896 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
22898 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
22899 (make_gcov_file_name): Use the canonical path name for generating
22901 (read_line): Fix handling of files with ascii null bytes.
22903 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
22905 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
22906 to initialise a vector register instead
22907 of using a const_int.
22909 2017-03-30 Jakub Jelinek <jakub@redhat.com>
22911 PR translation/80189
22912 * gimplify.c (omp_default_clause): Use %qs instead of %s in
22913 diagnostic messages.
22915 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
22918 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
22919 (dfp_diex_<mode>): Update mode of operand 1.
22920 * doc/extend.texi (dxex, dxexq): Document change to return type.
22921 (diex, diexq): Document change to argument type.
22923 2017-03-30 Martin Jambor <mjambor@suse.cz>
22926 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
22927 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
22928 it reflects the signature changes performed at the callee side.
22929 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
22930 to cgraph_build_function_type_skip_args.
22931 (build_function_decl_skip_args): Adjust call to the above function.
22933 2017-03-30 Jakub Jelinek <jakub@redhat.com>
22936 * config/i386/sse.md
22937 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
22938 register as dest whenever it is a MEM not rtx_equal_p to the
22939 corresponding dup operand, and when forcing into reg move the
22940 reg into the memory afterwards.
22941 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
22942 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
22943 for the force_reg mode.
22944 (avx512vl_vextractf128<mode>): Use register as dest either
22945 always when a MEM, or when it is a MEM not rtx_equal_p to the
22946 corresponding dup operand, or even not when it is a CONST_VECTOR
22947 depending on the mode and lo vs. hi.
22948 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
22950 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
22951 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
22952 Likewise. Require that operands[2] is even.
22953 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
22954 Remove extraneous parens. Require that operands[2] is a multiple
22956 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
22957 operands[0] is a MEM if <mask_applied>, the predicates/constraints
22958 disallow memory then.
22960 2017-03-30 Richard Biener <rguenther@suse.de>
22962 PR tree-optimization/77498
22963 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
22964 to non-constants over backedges.
22966 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
22968 PR rtl-optimization/80233
22969 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
22970 as last_combined_insn. Do not test for BARRIER_P separately.
22972 2017-03-29 Andreas Schwab <schwab@suse.de>
22975 * calls.c (prepare_call_address): Convert funexp to Pmode before
22976 copying to temp reg.
22978 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22980 PR tree-optimization/80158
22981 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
22982 Handle possible future case of more than one alternate
22984 (replace_rhs_if_not_dup): Likewise.
22985 (replace_one_candidate): Likewise.
22987 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
22989 PR rtl-optimization/80193
22990 * ira.c (ira): Do not check allocation for LRA.
22992 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
22994 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
22995 (nvptx_output_simt_exit): Declare.
22996 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
22997 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
22998 (init_softstack_frame): Move initialization of crtl->is_leaf to...
22999 (nvptx_declare_function_name): ...here. Emit declaration of local
23000 memory space buffer for omp_simt_enter insn.
23001 (nvptx_output_unisimt_switch): New.
23002 (nvptx_output_softstack_switch): New.
23003 (nvptx_output_simt_enter): New.
23004 (nvptx_output_simt_exit): New.
23005 * config/nvptx/nvptx.h (struct machine_function): New fields
23006 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
23007 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
23008 (UNSPECV_SIMT_EXIT): Ditto.
23009 (omp_simt_enter_insn): New insn.
23010 (omp_simt_enter): New expansion.
23011 (omp_simt_exit): New insn.
23012 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
23014 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
23015 (expand_GOMP_SIMT_ENTER_ALLOC): New.
23016 (expand_GOMP_SIMT_EXIT): New.
23017 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
23018 (GOMP_SIMT_ENTER_ALLOC): Ditto.
23019 (GOMP_SIMT_EXIT): Ditto.
23020 * target-insns.def (omp_simt_enter): New insn.
23021 (omp_simt_exit): Ditto.
23022 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
23024 (lower_rec_simd_input_clauses): Implement SIMT privatization.
23025 (lower_rec_input_clauses): Likewise.
23026 (lower_lastprivate_clauses): Handle SIMT privatization.
23028 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
23029 (ompdevlow_adjust_simt_enter): New.
23030 (find_simtpriv_var_op): New.
23031 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
23032 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
23034 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
23035 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
23036 (copy_decl_for_dup_finish): Ditto.
23038 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
23040 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
23043 * config/i386/i386.c (ix86_option_override_internal): Always
23044 allow -mpreferred-stack-boundary=3 for 64-bit targets.
23046 2017-03-28 Bin Cheng <bin.cheng@arm.com>
23048 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
23050 2017-03-28 Bin Cheng <bin.cheng@arm.com>
23052 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
23053 mark new edge's irreducible flag accordign to it.
23054 (vect_do_peeling): Check loop preheader edge's irreducible flag
23055 and pass it to function slpeel_add_loop_guard.
23057 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
23059 PR tree-optimization/80218
23060 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
23061 Update block frequencies and counts.
23063 2017-03-28 Richard Biener <rguenther@suse.de>
23065 PR tree-optimization/78644
23066 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
23067 of a simplification result we may not use it at all.
23069 2017-03-28 Richard Biener <rguenther@suse.de>
23072 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
23073 without arguments, generate default definition of a SSA name.
23075 2017-03-28 Richard Biener <rguenther@suse.de>
23077 PR middle-end/80222
23078 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
23079 TYPE_REF_CAN_ALIAS_ALL references.
23080 * fold-const.c (fold_indirect_ref_1): Likewise.
23082 2017-03-28 Martin Liska <mliska@suse.cz>
23085 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
23086 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
23088 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
23089 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
23091 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
23092 (EXTRA_SPECS): Define.
23093 (SUBTARGET_EXTRA_SPECS): Likewise.
23094 (SUBTARGET_CPP_SPEC): Likewise.
23095 * config/arc/elf.h (EXTRA_SPECS): Renamed to
23096 SUBTARGET_EXTRA_SPECS.
23097 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
23099 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
23101 * config/arc/simdext.md (vst64_insn): Update pattern.
23102 (vld32wh_insn): Likewise.
23103 (vld32wl_insn): Likewise.
23104 (vld64_insn): Likewise.
23105 (vld32_insn): Likewise.
23107 2017-03-28 Marek Polacek <polacek@redhat.com>
23110 * fold-const.c (fold_comparison): Use protected_set_expr_location
23111 instead of SET_EXPR_LOCATION.
23113 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
23115 * tree.c (add_expr): Avoid name lookup warning.
23117 2017-03-27 Jeff Law <law@redhat.com>
23119 PR tree-optimization/80216
23120 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
23121 function name. Limit recursion depth.
23122 (record_temporary_equivalences): Corresponding changes.
23124 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
23126 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
23129 2017-03-27 Jakub Jelinek <jakub@redhat.com>
23132 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
23134 * cfgcleanup.c (reg_note_cfa_p): New array.
23135 (insns_have_identical_cfa_notes): New function.
23136 (old_insns_match_p): Don't cross-jump in between /f
23137 and non-/f instructions. If both i1 and i2 are frame related,
23138 verify all CFA notes, their order and content.
23140 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
23143 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
23144 HImode and SImode with zero extend to DImode to one insn.
23145 (bswap<mode>2_extenddi): Likewise.
23146 (bswapsi2_extenddi): Likewise.
23147 (bswaphi2_extendsi): Likewise.
23148 (bswaphi2): Combine bswap HImode and SImode into one insn.
23149 Separate memory insns from swapping register.
23150 (bswapsi2): Likewise.
23151 (bswap<mode>2): Likewise.
23152 (bswaphi2_internal): Delete, no longer used.
23153 (bswapsi2_internal): Likewise.
23154 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
23155 store, and gpr<-gpr swap insns.
23156 (bswap<mode>2_store): Likewise.
23157 (bswaphi2_reg): Register only splitter, combine with the splitter.
23158 (bswaphi2 splitter): Likewise.
23159 (bswapsi2_reg): Likewise.
23160 (bswapsi2 splitter): Likewise.
23161 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
23162 the insns into load, store, and register/register insns.
23163 (bswapdi2_ldbrx): Likewise.
23164 (bswapdi2_load): Likewise.
23165 (bswapdi2_store): Likewise.
23166 (bswapdi2_reg): Likewise.
23168 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
23170 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
23171 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
23173 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
23176 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
23178 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
23179 special handling for target option conflicts between dform
23180 options (-mpower9-dform, -mpower9-dform-vector,
23181 -mpower9-dform-scalar) and -mno-direct-move.
23183 2017-03-27 Richard Biener <rguenther@suse.de>
23185 PR tree-optimization/80181
23186 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
23188 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
23190 * config/arc/predicates.md (move_double_src_operand): Replace the
23191 call to move_double_src_operand with a call to address_operand.
23193 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
23195 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
23196 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
23197 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
23199 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
23201 * config/arc/predicates.md (long_immediate_loadstore_operand):
23202 Consider scaled addresses cases.
23204 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
23206 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
23207 restored when in interrupt.
23208 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
23209 doesn't have delay slot.
23211 2017-03-27 Richard Biener <rguenther@suse.de>
23214 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
23215 inlined thunk clones.
23217 2017-03-27 Jakub Jelinek <jakub@redhat.com>
23220 * asan.c (instrument_derefs): Copy over last operand from
23221 original COMPONENT_REF to the new COMPONENT_REF with
23222 DECL_BIT_FIELD_REPRESENTATIVE.
23223 * ubsan.c (instrument_object_size): Likewise.
23225 2017-03-27 Richard Biener <rguenther@suse.de>
23227 PR tree-optimization/80170
23228 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
23229 sure DR/SCEV didnt fold in constants we do not see when looking
23230 at the reference base alignment.
23232 2017-03-27 Richard Biener <rguenther@suse.de>
23234 PR middle-end/80171
23235 * gimple-fold.c (fold_ctor_reference): Properly guard against
23236 NULL return value from canonicalize_constructor_val.
23238 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
23241 * config/i386/i386.c (ix86_expand_builtin)
23242 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
23243 flags reg setting and flags reg using instructions.
23244 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
23245 clobbering instructions to zero extend op2.
23247 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
23249 * doc/install.texi (Configuration) <--with-aix-soname>:
23250 Update link to AIX ld.
23252 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
23254 PR rtl-optimization/80160
23255 PR rtl-optimization/80159
23256 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
23257 reg_alternate_class into account.
23259 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
23262 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
23263 to consider in curr_insn_transform.
23265 2017-03-24 Jakub Jelinek <jakub@redhat.com>
23267 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
23268 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
23269 and emit_mode_inner.
23271 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23273 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
23274 argument to the overloaded builtin variants. Use the new flag to
23275 deprecate certain builtin variants.
23276 * config/s390/s390-builtin-types.def: Add new builtin types.
23277 * config/s390/s390-builtins.h: Support new flags field for
23278 overloaded builtins.
23279 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
23280 (s390_macro_to_expand): Enable vector float data type.
23281 (s390_cpu_cpp_builtins_internal): Indicate support of the new
23282 builtins by incrementing the __VEC__ version number.
23283 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
23285 (s390_resolve_overloaded_builtin): Emit error messages depending
23286 on the builtin flags.
23287 * config/s390/s390.c (s390_expand_builtin): Support additional
23288 flags argument. Change error message to match the messages
23289 emitted in s390-c.c.
23290 * config/s390/s390.md: New UNSPEC_* constants.
23291 (op_type): Add new instruction types.
23292 * config/s390/vecintrin.h: Add new builtins and test data class
23294 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
23295 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
23296 (VEC_INEXACT, VEC_NOINEXACT): New constants.
23297 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
23298 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
23299 ("vec_mergel<mode>"): V_HW -> VEC_HW.
23301 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
23302 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
23303 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
23304 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
23306 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
23307 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
23308 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
23309 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
23311 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
23312 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
23313 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
23314 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
23315 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
23316 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
23317 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
23319 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
23320 ("vec_scatter_element<V_HW_4:mode>_DI")
23321 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
23322 ("vec_fpint<mode>", "vflls")
23323 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
23324 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
23325 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
23326 ("*vec_cmphe<mode>_cc"): ... these.
23328 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
23329 mode constant instead of magic value.
23331 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23333 * config/s390/s390.c (s390_expand_vec_compare): Support other
23334 vector floating point modes than just V2DF.
23335 (s390_expand_vcond): Likewise.
23336 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
23337 (s390_cannot_change_mode_class): Prevent mode changes between TF
23338 and V1TF in vector registers.
23339 * config/s390/s390.md (DF, SF): New mode attributes.
23340 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
23341 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
23342 SFmode support for VRs.
23343 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
23345 (VFT, VF_HW): New mode iterators.
23346 (vw, sdx): New mode attributes.
23347 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
23348 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
23349 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
23350 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
23351 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
23352 also the new vector floating point modes. Renaming to ...
23354 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
23355 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
23356 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
23357 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
23358 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
23359 ("vec_unordered<mode>"): ... these.
23361 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
23362 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
23363 ("*vec_extendv2df"): New insn definitions.
23365 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23367 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
23368 ("mulditi3_2", "*muldi3_sign"): New patterns.
23369 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
23370 rename the pattern definition.
23372 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23374 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
23376 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
23378 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23380 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
23381 instruction if possible.
23382 * config/s390/vector.md (vec_halfnumelts): New mode
23384 ("*vec_vllezlf<mode>"): New pattern.
23386 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23388 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
23389 ("popcountv4si2", "popcountv2di2"): Rename to ...
23390 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
23391 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
23393 ("popcount<mode>2_vxe"): New pattern.
23395 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23397 * common/config/s390/s390-common.c (processor_flags_table): Add
23399 * config.gcc: Add arch12.
23400 * config/s390/driver-native.c (s390_host_detect_local_cpu):
23401 Default to arch12 for unknown CPU model numbers.
23402 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
23403 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
23404 PROCESSOR_max sanity check.
23405 * config/s390/s390-opts.h (enum processor_type): Add
23407 * config/s390/s390.c (processor_table): Add arch12.
23408 (s390_expand_builtin): Add check for B_VXE flag.
23409 (s390_issue_rate): Add PROCESSOR_ARCH12.
23410 (s390_get_sched_attrmask): Likewise.
23411 (s390_get_unit_mask): Likewise.
23412 (s390_sched_score): Enable z13 scheduling for arch12.
23413 (s390_sched_reorder): Likewise.
23414 (s390_sched_variable_issue): Likewise.
23415 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
23417 (s390_tune_attr): Use z13 scheduling also for arch12.
23418 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
23419 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
23420 (TARGET_VXE_P): New macros.
23421 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
23422 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
23423 * config/s390/s390.opt: Add arch12 as processor_type.
23425 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23427 * config/s390/s390.md
23428 ("fixuns_truncdddi2", "fixuns_trunctddi2")
23429 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
23430 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
23432 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
23433 Rename expanders to ...
23435 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
23436 ("fixuns_truncdddi2_emu"): ... these.
23438 ("fixuns_trunc<mode>si2_emu"): New expander.
23440 ("*fixuns_truncdfdi2_z13"): Rename to ...
23441 ("*fixuns_truncdfdi2_vx"): ... this.
23443 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23445 * config/s390/2964.md: Remove the single element vector compare
23446 instructions which are no longer used.
23447 * config/s390/s390.c (s390_select_ccmode): Remove handling of
23449 (s390_canonicalize_comparison): Remove handling of DFmode
23451 (s390_expand_vec_compare_scalar): Remove function.
23452 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
23453 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
23455 ("*cmp<mode>_ccs"): Add wfcdb instruction.
23457 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23459 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
23461 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
23462 will anyway by matched by mov<mode>_64dfp.
23464 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23466 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
23467 vlef/vstef. Add missing operand to vleif.
23469 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23471 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
23472 pair for all vector types with 64 bit elements.
23473 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
23474 * config/s390/vector.md (V_HW_64): ... here.
23475 (V_128_NOSINGLE): New mode iterator.
23476 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
23477 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
23478 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
23479 ("*vec_load_pairv2di"): Change to ...
23480 ("*vec_load_pair<mode>"): ... this one.
23482 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23484 * config/s390/constraints.md: Add comments.
23485 (jKK): Reject element sizes > 8 bytes.
23486 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
23488 * config/s390/s390.md: Add the s_operand checks formerly in
23489 s390_split_ok_p to various splitters where they are still
23491 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
23492 for 128 bit vectors. Plus two splitters.
23494 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23496 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
23499 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23502 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
23503 error if the boundary argument is not constant.
23505 2017-03-24 Jakub Jelinek <jakub@redhat.com>
23507 PR rtl-optimization/80112
23508 * loop-doloop.c (doloop_condition_get): Don't check condition
23509 if cmp isn't SET with IF_THEN_ELSE src.
23511 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23513 PR tree-optimization/80158
23514 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
23515 replacing a candidate statement, also replace it for the
23516 candidate's alternate interpretation.
23517 (replace_rhs_if_not_dup): Likewise.
23518 (replace_one_candidate): Likewise.
23520 2017-03-24 Richard Biener <rguenther@suse.de>
23522 PR tree-optimization/80167
23523 * graphite-isl-ast-to-gimple.c
23524 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
23526 (translate_isl_ast_to_gimple::get_rename): Likewise.
23528 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
23530 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
23531 handling of certain combinations of target options, including the
23532 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
23533 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
23535 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23538 * config/arm/arm.md (*load_multiple): Add reload_completed to
23539 matching condition.
23541 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23542 Richard Biener <rguenth@suse.de>
23544 PR tree-optimization/79908
23545 PR tree-optimization/80136
23546 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
23547 been cast away, gimplify_and_add suffices.
23549 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
23551 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
23553 2017-03-23 Richard Biener <rguenther@suse.de>
23555 PR tree-optimization/80032
23556 * gimplify.c (gimple_push_cleanup): Forced unconditional
23557 cleanups still have to go to the conditional_cleanups
23560 2017-03-22 Jakub Jelinek <jakub@redhat.com>
23562 PR tree-optimization/80072
23563 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
23565 (next_operand_entry_id): Change type to unsigned int.
23566 (sort_by_operand_rank): Make sure to return the right return value
23567 even if unsigned fields are bigger than INT_MAX.
23568 (struct oecount): Change cnt and id type to unsigned int.
23569 (oecount_hasher::equal): Formatting fix.
23570 (oecount_cmp): Make sure to return the right return value
23571 even if unsigned fields are bigger than INT_MAX.
23572 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
23575 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
23576 TREE_READONLY on result if writing it more than once.
23579 * doc/invoke.texi (-fsanitize=thread): Document that with
23580 -fnon-call-exceptions atomics are not able to throw
23584 * tsan.c: Include tree-eh.h.
23585 (instrument_builtin_call): Call maybe_clean_eh_stmt or
23586 maybe_clean_or_replace_eh_stmt where needed.
23587 (instrument_memory_accesses): Add cfg_changed argument.
23588 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
23589 if it returned true.
23590 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
23592 PR rtl-optimization/63191
23593 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
23594 wrapper function, moved the whole old content into ...
23595 (ix86_delegitimize_address_1): ... this. New inline function.
23596 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
23597 true as last argument instead of ix86_delegitimize_address.
23599 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
23601 * config/aarch64/aarch64.c (generic_branch_cost): Copy
23602 cortexa57_branch_cost.
23604 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
23606 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
23608 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23611 * doc/md.texi (Constraints): Document wA constraint.
23612 * config/rs6000/constraints.md (wA): New.
23613 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
23614 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
23615 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
23616 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
23618 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
23621 * gimplify.c (is_oacc_declared): New function.
23622 (oacc_default_clause): Use it to set default flags for acc declared
23623 variables inside parallel regions.
23624 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
23625 declared variables.
23626 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
23627 declare attribute to any decl as necessary.
23629 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
23632 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
23633 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
23634 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
23635 (arm_arch_lpae): This.
23636 * config/arm/arm.c (arm_arch7ve): Rename into ...
23637 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
23638 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
23641 2017-03-22 Martin Liska <mliska@suse.cz>
23644 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
23645 error message instead of an ICE.
23647 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23649 * doc/extend.texi (6.11 Additional Floating Types): Revise.
23651 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
23653 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
23655 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
23658 2017-03-21 Martin Sebor <msebor@redhat.com>
23660 * doc/extend.texi: Use "cannot" instead of "can't."
23661 * doc/hostconfig.texi: Same.
23662 * doc/install.texi: Same.
23663 * doc/invoke.texi: Same.
23664 * doc/loop.texi: Same.
23665 * doc/md.texi: Same.
23666 * doc/objc.texi: Same.
23667 * doc/rtl.texi: Same.
23668 * doc/tm.texi: Same.
23669 * doc/tm.texi.in: Same.
23670 * doc/trouble.texi: Same.
23672 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
23675 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
23676 (collect_checksum_attributes): Set it.
23677 (die_checksum_ordered): Use it.
23679 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23681 PR tree-optimization/79908
23682 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
23683 change: For a VA_ARG whose LHS has been cast away, use
23684 force_gimple_operand to construct the side effects.
23686 2017-03-21 David Malcolm <dmalcolm@redhat.com>
23688 PR translation/80001
23689 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
23690 more amenable to translation.
23691 (oacc_loop_auto_partitions): Likewise.
23693 2017-03-21 Marek Polacek <polacek@redhat.com>
23694 Martin Sebor <msebor@redhat.com>
23696 PR tree-optimization/80109
23697 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
23698 on INTEGRAL_TYPE_P.
23700 2017-03-21 Jakub Jelinek <jakub@redhat.com>
23701 Segher Boessenkool <segher@kernel.crashing.org>
23704 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
23705 check reg_used_between_p between insn and one of succ or succ2
23706 depending on if succ is artificial insn not inserted into insn
23709 2017-03-21 Martin Liska <mliska@suse.cz>
23711 PR gcov-profile/80081
23712 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
23713 * doc/gcc.texi: Include gcov-dump stuff.
23714 * doc/gcov-dump.texi: New file.
23716 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
23718 PR rtl-optimization/79150
23719 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
23720 conditional jump, if the jump is the last insn of the loop.
23722 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23723 Richard Biener <rguenth@suse.de>
23725 PR tree-optimization/79908
23726 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
23727 been cast away, use force_gimple_operand to construct the side
23730 2017-03-21 Martin Liska <mliska@suse.cz>
23732 PR libfortran/79956
23733 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
23736 2017-03-21 Brad Spengler <spender@grsecurity.net>
23739 * plugin.c (htab_hash_plugin): New function.
23740 (add_new_plugin): Use it and adjust.
23741 (parse_plugin_arg_opt): Adjust.
23742 (init_one_plugin): Likewise.
23744 2017-03-21 Richard Biener <rguenther@suse.de>
23746 PR tree-optimization/80032
23747 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
23748 if set force the cleanup to happen unconditionally.
23749 (gimplify_target_expr): Push inserted clobbers with force_uncond
23750 to avoid them being removed by control-dependent DCE.
23752 2017-03-21 Richard Biener <rguenther@suse.de>
23754 PR tree-optimization/80122
23755 * tree-inline.c (copy_bb): Do not expans va-arg packs or
23756 va_arg_pack_len when the inlined call stmt requires pack
23758 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
23760 2017-03-21 Jakub Jelinek <jakub@redhat.com>
23763 * tsan.c (instrument_builtin_call): If the memory model argument
23764 is not a constant, assume it is valid.
23767 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
23770 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
23772 PR rtl-optimization/79910
23773 * combine.c (can_combine_p): Do not allow combining an I0 or I1
23774 if its dest is used by an insn before I2 (other than the combined
23775 insns themselves, which are properly handled already).
23777 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
23780 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
23782 * combine.c (record_used_regs): New static function.
23783 (try_combine): Handle situations where there is an additional
23784 instruction between I2 and I3 which needs to have a LOG_LINK
23788 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
23790 * combine.c (try_combine): Delete redundant i1 test. Call
23791 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
23793 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23796 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
23797 alternatives 13/14.
23799 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23801 PR tree-optimization/80054
23802 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
23803 the optimization if a PHI or any of its arguments is not dominated
23804 by the candidate's basis. Use gphi* rather than gimple* as
23806 (replace_profitable_candidates): Clean up a gimple* variable that
23807 should be a gphi* variable.
23809 2017-03-20 Martin Sebor <msebor@redhat.com>
23812 * doc/extend.texi (attribute constructor): Document present limitation.
23814 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
23817 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
23818 __POWER9_VECTOR__ #ifdef control, change template definition to
23819 use Power9-specific built-in function.
23820 (vec_any_eq): Likewise.
23821 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
23822 to control outcomes from this test.
23823 (vector_ae_<mode>p): For VEC_F modes, likewise.
23825 2017-03-20 Ian Lance Taylor <iant@google.com>
23827 * config/i386/i386.c (ix86_function_regparm): Save an extra
23828 register for -fsplit-stack with DECL_STATIC_CHAIN.
23830 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
23833 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
23834 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
23836 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
23838 * config/riscv/riscv.c (riscv_print_operand): Use "fence
23840 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
23843 2017-03-20 Marek Polacek <polacek@redhat.com>
23846 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
23848 2017-03-20 Richard Biener <rguenther@suse.de>
23850 PR tree-optimization/80113
23851 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
23852 allocate extra SSA name for PHI def.
23853 (add_close_phis_to_outer_loops): Likewise.
23854 (add_close_phis_to_merge_points): Likewise.
23855 (copy_loop_close_phi_args): Likewise.
23856 (copy_cond_phi_nodes): Likewise.
23858 2017-03-20 Martin Liska <mliska@suse.cz>
23860 PR middle-end/79753
23861 * tree-chkp.c (chkp_build_returned_bound): Do not build
23862 returned bounds for a LHS that's not a BOUNDED_P type.
23864 2017-03-20 Martin Liska <mliska@suse.cz>
23868 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
23869 COMPLEX_CST and VECTOR_CST.
23871 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23874 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
23875 target operand. A new splitter adds the clobber statement in case
23876 the target operand is dead anyway.
23878 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
23880 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
23881 to age-old versions of binutils and glibc.
23883 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
23885 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
23887 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
23889 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
23891 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
23893 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
23894 requirement for binutils 2.13.
23896 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
23898 * combine.c (try_combine): Delete redundant i1 test. Call
23899 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
23901 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
23903 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
23904 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
23906 <riscv64-*-elf>: Re-arrange section
23907 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
23908 <riscv32-*-linux>: Likewise.
23909 <riscv64-*-elf>: Likewise
23910 <riscv64-*-linux>: Likewise.
23912 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
23915 * aarch64.opt(verbose-cost-dump): Fix typo.
23917 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
23920 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
23921 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
23923 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
23925 * reload.c (find_reloads): When reloading a nonoffsettable address,
23926 use RELOAD_OTHER for it and its address reloads.
23928 PR rtl-optimization/79910
23929 * combine.c (record_used_regs): New static function.
23930 (try_combine): Handle situations where there is an additional
23931 instruction between I2 and I3 which needs to have a LOG_LINK
23934 2017-03-17 Jeff Law <law@redhat.com>
23936 PR tree-optimization/71437
23937 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
23938 conditional in the hash table first.
23939 (vrp_dom_walker::before_dom_children): Extract condition from
23940 ASSERT_EXPR. Record condition, its inverion and any implied
23941 conditions as well.
23943 2017-03-17 Marek Polacek <polacek@redhat.com>
23944 Markus Trippelsdorf <markus@trippelsdorf.de>
23946 PR tree-optimization/80079
23947 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
23950 2017-03-17 Richard Biener <rguenther@suse.de>
23952 PR middle-end/80075
23953 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
23954 Properly verify the LHS before the RHS possibly claims to be
23956 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
23959 2017-03-17 Martin Jambor <mjambor@suse.cz>
23961 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
23962 (List of -O2 options): Likewise.
23963 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
23966 2017-03-17 Tom de Vries <tom@codesourcery.com>
23968 * gcov-dump.c (print_usage): Print bug_report_url.
23970 2017-03-17 Richard Biener <rguenther@suse.de>
23972 PR middle-end/80050
23973 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
23974 (parser::peek): Likewise.
23976 2017-03-17 Richard Biener <rguenther@suse.de>
23978 PR tree-optimization/80048
23979 * sese.c (free_sese_info): Properly release rename_map and
23980 copied_bb_map elements.
23982 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
23984 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
23985 Add linked-list forward and backlinks. Insert on
23986 construction, remove on destruction.
23987 (class pass_store_merging): Add m_stores_head field.
23988 (pass_store_merging::terminate_and_process_all_chains):
23989 Iterate over m_stores_head list.
23990 (pass_store_merging::terminate_all_aliasing_chains):
23992 (pass_store_merging::execute): Check for debug stmts first.
23993 Push new chains onto the m_stores_head stack.
23995 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
23998 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
23999 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
24000 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
24002 2017-03-16 Jeff Law <law@redhat.com>
24004 PR tree-optimization/71437
24005 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
24006 member function. Implementation moved into after_dom_children
24007 member function and into the threader's thread_outgoing_edges
24009 (dom_opt_dom_walker::after_dom_children): Simplify by moving
24010 some code into new thread_outgoing_edges.
24011 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
24012 definition. Simplify marker handling (do it here). Assume we always
24013 have the available expression and the const/copies tables.
24014 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
24016 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
24017 * tree-vrp.c (equiv_stack): No longer file scoped.
24018 (vrp_dom_walker): New class.
24019 (vrp_dom_walker::before_dom_children): New member function.
24020 (vrp_dom_walker::after_dom_children): Likewise.
24021 (identify_jump_threads): Setup domwalker. Use it rather than
24022 walking edges in a random order by hand. Simplify setup/finalization.
24023 (finalize_jump_threads): Remove.
24024 (vrp_finalize): Do not call identify_jump_threads here.
24025 (execute_vrp): Do it here instead and call thread_through_all_blocks
24028 PR tree-optimization/71437
24029 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
24031 (simplify_stmt_for_jump_threading): Add basic_block argument. All
24033 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
24034 (dom_opt_dom_walker::thread_across_edge): Remove
24035 handle_dominating_asserts argument. All callers changed.
24036 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
24037 changes. Remove calls to lhs_of_dominating_assert. Other
24038 uses of handle_dominating_asserts turn into unconditional code
24039 (simplify_control_stmt_condition_1): Likewise.
24040 (simplify_control_stmt_condition): Likewise.
24041 (thread_through_normal_block, thread_across_edge): Likewise.
24042 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
24043 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
24044 object if it is not an SSA_NAME.
24045 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
24046 before calling into the VRP specific simplifiers.
24047 (identify_jump_threads): Remove handle_dominating_asserts
24050 2017-03-16 Jakub Jelinek <jakub@redhat.com>
24053 * tree-diagnostic.c (default_tree_printer): No longer static.
24054 * tree-diagnostic.h (default_tree_printer): New prototype.
24056 2017-03-16 Tamar Christina <tamar.christina@arm.com>
24058 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
24059 Change ins into fmov.
24061 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24063 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
24064 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
24065 Use h_con constraint for operand 1.
24066 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
24067 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
24069 2017-03-15 Jeff Law <law@redhat.com>
24071 PR tree-optimization/71437
24072 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
24073 (record_temporary_equivalences): Use it.
24075 PR tree-optimization/71437
24076 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
24077 tree-ssa-scopedtables.
24078 (lookup_avail_expr, build_and_record_new_cond): Likewise.
24079 (record_conditions, record_cond, vuse_eq): Likewise.
24080 (record_edge_info): Adjust to API tweak of record_conditions.
24081 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
24082 (record_temporary_equivalences, optimize_stmt): Likewise.
24083 (eliminate_redundant_computations): Likewise.
24084 (record_equivalences_from_stmt): Likewise.
24085 * tree-ssa-scopedtables.c: Include options.h and params.h.
24086 (vuse_eq): New function, moved from tree-ssa-dom.c
24087 (build_and_record_new_cond): Likewise.
24088 (record_conditions): Likewise. Accept vector of conditions rather
24089 than edge_equivalence structure for first argument.
24090 for the first argument.
24091 (avail_exprs_stack::lookup_avail_expr): New member function, moved
24092 from tree-ssa-dom.c.
24093 (avail_exprs_stack::record_cond): Likewise.
24094 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
24095 from tree-ssa-dom.c.
24096 (avail_exprs_stack): Add new member functions lookup_avail_expr
24098 (record_conditions): Declare.
24100 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
24103 * lra-constraints.c (process_alt_operands): Increase reject for
24104 reloading an input/output operand.
24106 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
24109 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
24110 insns to convert from signed/unsigned char/short to IEEE 128-bit
24112 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
24114 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
24117 * config/i386/i386.c (ix86_vector_duplicate_value): Create
24118 subreg of inner mode for values already in registers.
24120 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
24122 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
24123 iteration reg is used after the loop.
24125 2017-03-14 Martin Sebor <msebor@redhat.com>
24127 PR tree-optimization/79800
24128 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
24129 precision in negative-positive range.
24130 (format_floating): Call non-const overload with adjusted precision.
24132 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
24135 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
24138 2017-03-14 Martin Sebor <msebor@redhat.com>
24140 PR middle-end/80020
24141 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
24142 * builtins.def (aligned_alloc): Use it.
24145 * Makefile.in (GTFILES): Add calls.c.
24146 * calls.c: Include "gt-calls.h".
24148 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
24150 PR rtl-optimization/79728
24151 * regs.h (struct target_regs): New field
24152 x_contains_allocatable_regs_of_mode.
24153 (contains_allocatable_regs_of_mode): New macro.
24154 * reginfo.c (init_reg_sets_1): Initialize it, and change
24155 contains_reg_of_mode so it includes global regs as well.
24156 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
24157 rather than contains_regs_of_mode.
24159 2017-03-14 Martin Liska <mliska@suse.cz>
24161 * doc/invoke.texi: Document options that can't be combined with
24162 -fcheck-pointer-bounds.
24164 2017-03-14 Martin Liska <mliska@suse.cz>
24166 PR middle-end/79831
24167 * doc/invoke.texi (-Wchkp): Document the option.
24169 2017-03-14 Martin Liska <mliska@suse.cz>
24171 * Makefile.in: Install gcov-dump.
24173 2017-03-14 Martin Liska <mliska@suse.cz>
24175 * multiple_target.c (expand_target_clones): Bail out for
24176 an invalid attribute.
24178 2017-03-14 Richard Biener <rguenther@suse.de>
24180 * alias.c (struct alias_set_entry): Pack properly.
24181 * cfgloop.h (struct loop): Likewise.
24182 * cse.c (struct set): Likewise.
24183 * ipa-utils.c (struct searchc_env): Likewise.
24184 * loop-invariant.c (struct invariant): Likewise.
24185 * lra-remat.c (struct cand): Likewise.
24186 * recog.c (struct change_t): Likewise.
24187 * rtl.h (struct address_info): Likewise.
24188 * symbol-summary.h (function_summary): Likewise.
24189 * tree-loop-distribution.c (struct partition): Likewise.
24190 * tree-object-size.c (struct object_size_info): Likewise.
24191 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
24192 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
24193 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
24194 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
24195 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
24196 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
24197 (struct _stmt_vec_info): Likewise.
24199 2017-03-14 Martin Liska <mliska@suse.cz>
24202 * multiple_target.c (create_dispatcher_calls): Check that
24203 a target can create a function dispatcher.
24205 2017-03-14 Martin Liska <mliska@suse.cz>
24208 * multiple_target.c (expand_target_clones): Drop local.local
24209 flag for default implementation.
24211 2017-03-14 Richard Biener <rguenther@suse.de>
24213 PR tree-optimization/80030
24214 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
24216 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
24218 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
24219 gcc_fallthrough() instead of __attribute__((fallthrough));
24221 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
24223 * doc/gcc.texi: Remove "up" link to (DIR).
24224 * doc/gccint.texi: Ditto.
24226 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
24228 * doc/install.texi (Specific) <avr>: Remove reference to
24231 2017-03-13 Jeff Law <law@redhat.com>
24233 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
24234 attribute rather than comments.
24236 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
24237 match_scratch operand is highest.
24239 2017-03-13 Martin Liska <mliska@suse.cz>
24241 PR middle-end/78339
24242 * ipa-pure-const.c (warn_function_noreturn): If the declarations
24243 is a CHKP clone, use original declaration.
24245 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24247 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
24248 (arc_conditional_register_usage): Use a different allocation order
24249 when optimizing for size.
24250 * common/config/arc/arc-common.c (arc_option_optimization_table):
24251 Section anchors default on when optimizing for size.
24253 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24255 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
24257 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24259 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
24260 * config/arc/arc.md (cpu_facility): Add cd variant.
24261 (*movqi_insn): Add code density variant.
24262 (*movhi_insn): Likewise.
24263 (*movqi_insn): Likewise.
24264 (*addsi3_mixed): Likewise.
24265 (subsi3_insn): Likewise.
24267 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24269 * config/arc/arc.md (movsi_cond_exec): Update constraint.
24271 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24273 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
24274 expressions with MINUS and UNARY ops.
24276 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24279 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
24281 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
24282 between vec_select and vector argument.
24283 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
24284 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
24285 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
24286 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
24287 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
24288 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
24290 2017-03-13 Richard Biener <rguenther@suse.de>
24293 * params.def (vect-max-peeling-for-alignment): Fix typo.
24295 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
24297 * doc/install.texi (Specific) <mips-*-*>: Remove description of
24298 issue that only occurred with binutils below 2.18.
24300 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
24302 * doc/install.texi (Specific) <cris-axis-elf>: No longer
24303 refer to binutils 2.11/2.12 minimum.
24305 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
24307 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
24308 ftp.kernel.org and simplify binutils requirement.
24310 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
24312 * doc/invoke.texi (Warning Options): Fix spelling of link-time
24314 (Optimize Options): Ditto. Also remove redundancy.
24316 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24318 PR translation/79848
24319 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
24321 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
24322 to G_ to avoid double translation.
24324 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24326 PR translation/79923
24327 * auto-profile.c (get_combined_location): Convert leading
24328 character of diagnostics to lower case and remove trailing period.
24329 (read_profile): Likewise for various diagnostics.
24330 * config/arm/arm.c (arm_option_override): Remove trailing period
24331 from various diagnostics.
24332 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
24333 (msp430_expand_delay_cycles): Likewise.
24335 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24338 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
24339 full command-line argument, rather than just "str".
24340 (aarch64_validate_march): Likewise.
24341 (aarch64_validate_mtune): Likewise.
24343 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
24345 PR rtl-optimization/78911
24346 * lra-assigns.c (must_not_spill_p): New function.
24347 (spill_for): Use it.
24349 2017-03-10 Jakub Jelinek <jakub@redhat.com>
24351 PR tree-optimization/79981
24352 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
24353 ATOMIC_COMPARE_EXCHANGE ifn result.
24354 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
24355 IFN_ATOMIC_COMPARE_EXCHANGE.
24357 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24360 * opts.c (parse_sanitizer_options): Add missing question mark to
24361 "did you mean" message.
24363 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24365 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
24367 (VMULEUH_UNS): Likewise.
24368 (VMULOUB_UNS): Likewise.
24369 (VMULOUH_UNS): Likewise.
24370 * config/rs6000/rs6000.c (builtin_function_type): Remove
24371 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
24373 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24376 * read-rtl-function.c (function_reader::read_rtx_operand): Update
24377 x with result of extra_parsing_for_operand_code_0.
24378 (function_reader::extra_parsing_for_operand_code_0): Convert
24379 return type from void to rtx, returning x. When reading
24380 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
24381 larger size containing struct block_symbol.
24383 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
24385 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
24386 -mfloat128-hardware without -m64.
24388 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
24391 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
24392 entries to the case statement that marks unsigned arguments to
24393 overloaded functions.
24395 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
24397 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
24398 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
24400 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
24403 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
24404 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
24406 2017-03-10 Martin Liska <mliska@suse.cz>
24410 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
24411 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
24414 2017-03-10 Olivier Hainque <hainque@adacore.com>
24416 * tree-switch-conversion (array_value_type): Start by resetting
24417 candidate type to it's main variant.
24419 2017-03-10 Jakub Jelinek <jakub@redhat.com>
24421 PR rtl-optimization/79909
24422 * combine.c (try_combine): Use simplify_replace_rtx on individual
24423 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
24424 of the whole CALL_INSN_FUNCTION_USAGE.
24426 PR tree-optimization/79972
24427 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
24428 get_range_info on SSA_NAMEs. Formatting fixes.
24430 2017-03-10 Richard Biener <rguenther@suse.de>
24431 Jakub Jelinek <jakub@redhat.com>
24433 PR tree-optimization/77975
24434 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
24435 edge to be constant.
24436 (get_val_for): For constant x return it. Formatting fix.
24437 (loop_niter_by_eval): Avoid pointless looping if the next iteration
24438 would use the same bases as the current one.
24440 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24442 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
24443 instead of vec_select for V1TImode.
24444 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
24446 (VSX_LE_128): Add V1TI to this mode iterator.
24447 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
24448 (*vsx_le_perm_store_<mode>): Likewise.
24449 (pre-reload splitter for VSX stores): Likewise.
24450 (post-reload splitter for VSX stores): Likewise.
24451 (*vsx_xxpermdi2_le_<mode>): Likewise.
24452 (*vsx_lxvd2x2_le_<mode>): Likewise.
24453 (*vsx_stxvd2x2_le_<mode>): Likewise.
24455 2017-03-09 Michael Eager <eager@eagercon.com>
24457 Correct failures with --enable-checking=yes,rtl.
24459 * config/microblaze/microblaze.c (microblaze_expand_shift):
24460 Replace GET_CODE test with CONST_INT_P and INTVAL test with
24461 test for const0_rtx.
24462 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
24463 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
24465 2017-03-09 Richard Biener <rguenther@suse.de>
24467 PR tree-optimization/79977
24468 * graphite-scop-detection.c (scop_detection::merge_sese):
24469 Handle the case of extra exits to blocks dominating the entry.
24471 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
24473 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
24476 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
24478 PR rtl-optimization/79949
24479 * lra-constraints.c (process_alt_operands): Check memory when
24480 trying to predict a cycle. Print about the overall increase.
24482 2017-03-09 Richard Biener <rguenther@suse.de>
24484 PR middle-end/79971
24485 * gimple-expr.c (useless_type_conversion_p): Preserve
24486 TYPE_SATURATING for fixed-point types.
24488 2017-03-09 Richard Biener <rguenther@suse.de>
24491 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
24492 alignment of BLKmode params.
24494 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24497 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
24498 (VALL_NO_V2Q): Likewise.
24500 * config/aarch64/aarch64-simd.md
24501 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
24503 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
24504 VALL_NO_V2Q mode iterator.
24505 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
24507 2017-03-09 Martin Liska <mliska@suse.cz>
24509 PR tree-optimization/79631
24510 * tree-chkp-opt.c (chkp_is_constant_addr): Call
24511 tree_int_cst_sign_bit just for INTEGER constants.
24513 2017-03-09 Martin Liska <mliska@suse.cz>
24517 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
24520 2017-03-09 Marek Polacek <polacek@redhat.com>
24523 * tree.c (inchash::add_expr): Handle TREE_VEC.
24525 2017-03-09 Martin Liska <mliska@suse.cz>
24528 (chkp_narrow_size_and_offset): New function.
24529 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
24530 (void chkp_parse_bit_field_ref): New function.
24531 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
24532 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
24534 2017-03-09 Martin Liska <mliska@suse.cz>
24537 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
24538 (chkp_find_bounds_1): Remove gcc_unreachable.
24540 2017-03-09 Jakub Jelinek <jakub@redhat.com>
24543 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
24544 BUILT_IN_SYNC*, determine the access type from the size suffix and
24545 always build a MEM_REF with that type. Handle forgotten
24546 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
24549 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
24550 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
24551 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
24552 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
24553 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
24554 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
24555 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
24556 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
24557 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
24558 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
24559 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
24560 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
24561 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
24562 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
24563 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
24564 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
24565 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
24566 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
24567 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
24568 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
24569 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
24570 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
24571 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
24572 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
24573 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
24574 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
24575 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
24576 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
24577 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
24578 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
24579 definitions outside of __OPTIMIZE__ guarded section.
24582 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
24583 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
24584 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
24585 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
24588 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24590 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
24591 ("vfenez<mode>"): Add missing constraints.
24593 2017-03-08 Martin Sebor <msebor@redhat.com>
24596 * config/nds32/nds32.c (nds32_option_override):
24597 Fix misspelled diagnostic.
24599 2017-03-08 Jakub Jelinek <jakub@redhat.com>
24602 * gimplify.c (gimplify_omp_for): Replace index var in outer
24603 taskloop statement with an artificial variable and add
24604 OMP_CLAUSE_PRIVATE clause for it.
24606 2017-03-08 Richard Biener <rguenther@suse.de>
24608 PR tree-optimization/79955
24609 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
24610 for accesses that are completely outside of the variable.
24612 2017-03-08 Andrew Haley <aph@redhat.com>
24614 PR tree-optimization/79943
24615 * tree-ssa-loop-split.c (compute_new_first_bound): When
24616 calculating the new upper bound, (END-BEG) should be added, not
24619 2017-03-08 Jakub Jelinek <jakub@redhat.com>
24621 * config/avr/avr.md (setmemhi): Make sure match_dup
24622 operand number comes before match_scratch.
24624 2017-03-08 Richard Biener <rguenther@suse.de>
24626 PR tree-optimization/79920
24627 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
24628 with ncopies == 1 to ...
24629 (vect_transform_slp_perm_load): ... here. Properly compute
24630 all element loads by iterating VF times over the group. Do
24631 not handle ncopies (computed in a broken way) in
24632 vect_create_mask_and_perm.
24634 2017-03-08 Jakub Jelinek <jakub@redhat.com>
24637 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
24638 is a uniform vector, use uniform_vector_p return value instead of
24639 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
24641 2017-03-07 Marek Polacek <polacek@redhat.com>
24643 PR middle-end/79809
24644 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
24645 (alloca_call_type): Likewise.
24647 2017-03-07 Martin Liska <mliska@suse.cz>
24649 * gcov.c (process_args): Put comment to correct location.
24651 2017-03-07 Martin Liska <mliska@suse.cz>
24653 PR middle-end/68270
24654 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
24655 Use array_at_struct_end_p instead of DECL_CHAIN (field).
24656 (chkp_narrow_bounds_for_field): Likewise.
24657 (chkp_parse_array_and_component_ref): Pass one more argument to
24660 2017-03-07 Richard Biener <rguenther@suse.de>
24662 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
24665 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
24667 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
24668 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
24670 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24673 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
24674 to end of description.
24675 (PARAM_MAX_STORES_TO_MERGE): Likewise.
24677 2017-03-07 Jakub Jelinek <jakub@redhat.com>
24679 PR rtl-optimization/79901
24680 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
24682 (*avx512f_<code><mode>3<mask_name>): ... this.
24683 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
24684 iterator instead of VI8_AVX2_AVX512BW.
24686 PR rtl-optimization/79901
24687 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
24688 min/max expander, expand it using expand_vec_cond_expr.
24691 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
24694 2017-03-06 Jakub Jelinek <jakub@redhat.com>
24697 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
24698 to void * for PCH reasons.
24699 * dwarf2out.c (output_loc_operands, output_die): Cast
24700 v.val_vec.array to unsigned char *.
24702 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
24705 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
24708 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
24710 PR rtl-optimization/79571
24711 * lra-constraints.c (process_alt_operands): Calculate static
24712 reject and subtract it from overall when only addresses will be
24715 2017-03-06 Julia Koval <julia.koval@intel.com>
24718 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
24719 incoming stack boundary to 128 for 64-bit targets.
24721 2017-03-06 Richard Biener <rguenther@suse.de>
24723 PR tree-optimization/79894
24724 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
24725 to NULL after folding it.
24727 2017-03-06 Richard Biener <rguenther@suse.de>
24729 PR tree-optimization/79824
24730 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
24731 check disabling peeling for gaps.
24733 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
24735 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
24736 attributes): Document gettimeofday.
24738 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
24740 * config/s390/s390.c (s390_option_override_internal): Set
24741 PARAM_MIN_VECT_LOOP_BOUND
24743 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
24745 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
24746 * config/s390/s390.md: Likewise.
24748 2017-03-06 Jakub Jelinek <jakub@redhat.com>
24751 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
24752 (<avx2_avx512>_perm<mode>): Rename to ...
24753 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
24755 (<avx512>_perm<mode>_mask): Rename to ...
24756 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
24758 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
24759 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
24760 instead of VI8F_256_512.
24761 (avx512f_perm<mode>): New define_expand.
24762 (avx512f_perm<mode>_mask): Likewise.
24763 (avx512f_perm<mode>_1<mask_name>): New define_insn.
24764 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
24766 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
24768 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
24769 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
24771 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
24773 2017-03-06 Martin Liska <mliska@suse.cz>
24776 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
24777 when having a SSA NAME w/o VAR_DECL assigned to it.
24779 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
24781 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
24782 msa_dpsub_<su>_d): Fix MODE for vec_select.
24784 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
24786 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
24788 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
24790 2017-03-06 Richard Biener <rguenther@suse.de>
24792 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
24793 * plugin.c (register_plugin_info): Likewise.
24794 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
24796 2017-03-05 Jakub Jelinek <jakub@redhat.com>
24798 * config/i386/sse.md (sse_storehps, sse_storelps,
24799 avx_<castmode><avxsizesuffix>_<castmode>,
24800 avx512f_<castmode><avxsizesuffix>_<castmode>,
24801 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
24802 in condition that at least one operand is not a MEM.
24804 2017-03-03 Jakub Jelinek <jakub@redhat.com>
24806 PR middle-end/79805
24807 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
24808 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
24810 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
24811 gimple_call_nothrow_p flag based on whether original builtin can throw.
24812 If it can, emit following stmts on the fallthrough edge.
24813 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
24814 don't create new bb if inserting just debug stmts on the edge, try to
24815 insert them on the fallthru bb or just reset debug stmts.
24817 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
24820 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
24821 restore recog_data (including the operand rtxes inside it) around
24822 the call to get_insn_template.
24824 2017-03-03 Martin Sebor <msebor@redhat.com>
24826 PR tree-optimization/79699
24827 * context.c (context::~context): Free MPFR caches to avoid
24828 a memory leak on program exit.
24830 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24832 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
24833 Use wide_int::ulow () instead of .elt (0).
24835 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
24837 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
24838 (*pushxf): Limit oF constraint to 32bit targets and add oC
24839 constraint for 64bit targets.
24840 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
24841 (*pushdf): Change rmF constraint to rmC.
24843 2017-03-03 Martin Liska <mliska@suse.cz>
24845 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
24846 Remove unused variable.
24848 2017-03-03 Jakub Jelinek <jakub@redhat.com>
24851 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
24852 is a memory operand, increase num_memory.
24853 (ix86_expand_args_builtin): Likewise.
24855 2017-03-03 Jan Hubicka <jh@suse.cz>
24858 * ipa-devirt.c (maybe_record_node): Properly handle
24859 __cxa_pure_virtual visibility.
24861 2017-03-03 Martin Liska <mliska@suse.cz>
24863 PR tree-optimization/79803
24864 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
24866 (pass_loop_prefetch::execute): Disabled optimization if an
24867 assumption about L1 cache size is not met.
24869 2017-03-03 Martin Liska <mliska@suse.cz>
24871 PR rtl-optimization/79574
24872 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
24873 (hash_scan_set): Likewise.
24874 (dump_hash_table): Likewise.
24875 (hoist_code): Likewise.
24877 2017-03-03 Richard Biener <rguenther@suse.de>
24879 * fixed-value.c (fixed_from_string): Restore use of elt (1)
24880 in place of uhigh ().
24881 (fixed_convert_from_real): Likewise.
24883 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
24886 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
24888 2017-03-03 Richard Biener <rguenther@suse.de>
24890 PR middle-end/79818
24891 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
24892 TYPE_OVERFLOW_UNDEFINED check.
24894 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24896 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
24898 (vector_ae_<mode>_p): Likewise.
24899 (vector_nez_<mode>_p): Likewise.
24900 (vector_ne_v2di_p): Likewise.
24901 (vector_ae_v2di_p): Likewise.
24902 (vector_ne_<mode>_p): Likewise.
24903 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
24905 (vsx_tsqrt<mode>2_fe): Likewise.
24907 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
24910 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
24912 2017-03-02 Jakub Jelinek <jakub@redhat.com>
24914 PR rtl-optimization/79780
24915 * cprop.c (one_cprop_pass): When second and further conditional trap
24916 in a single basic block is turned into an unconditional trap, turn it
24917 into a deleted note to avoid RTL verification failures.
24919 2017-03-02 Richard Biener <rguenther@suse.de>
24921 * fold-const.c (const_binop): Use ulow () instead of elt (0).
24923 2017-03-02 Richard Biener <rguenther@suse.de>
24925 PR tree-optimization/79345
24927 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
24928 param and abort the walk, returning -1 if it is hit.
24929 (walk_aliased_vdefs): Take a limit param and pass it on.
24930 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
24931 defaulting to 0 and return a signed int.
24932 * tree-ssa-uninit.c (struct check_defs_data): New struct.
24933 (check_defs): New helper.
24934 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
24935 about uninitialized memory.
24936 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
24937 bogus uninitialized warning.
24938 (fixed_convert_from_real): Likewise.
24940 2017-03-02 Bin Cheng <bin.cheng@arm.com>
24942 PR tree-optimization/66768
24943 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
24944 iv_use if base object can't be determined.
24946 2017-03-02 Jakub Jelinek <jakub@redhat.com>
24948 PR tree-optimization/79345
24949 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
24950 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
24951 (get_pattern_stats): Initialize it.
24952 * genemit.c (gen_expand): Verify match_scratch numbers come after
24953 match_operand/match_dup numbers.
24954 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
24955 match_scratch numbers.
24956 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
24958 * config/s390/s390.md (trunctdsd2): Likewise.
24960 2017-03-02 Richard Biener <rguenther@suse.de>
24962 * wide-int.h (wide_int_storage::operator=): Implement in terms
24965 2017-03-02 Richard Biener <rguenther@suse.de>
24967 PR tree-optimization/79777
24968 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
24969 the to insert expression to sth existing.
24971 2017-03-01 Martin Sebor <msebor@redhat.com>
24973 PR middle-end/79692
24974 * gimple-ssa-sprintf.c
24975 (directive::known_width_and_precision): New function.
24976 (format_integer): Use it.
24977 (get_mpfr_format_length): Consider the full range of precision
24978 when computing %g output with the # flag. Set the likely byte
24979 count to 3 rather than 1 when precision is indeterminate.
24980 (format_floating): Correct the lower bound of precision.
24982 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24984 * doc/invoke.texi: Document default code model for 64-bit Linux.
24986 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24989 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
24990 udiv rather than div since input pattern is unsigned.
24992 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
24994 * config/i386/i386.c (print_reg): Warn for values of
24995 unsupported size in integer register.
24997 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
25000 * config/rs6000/predicates.md (current_file_function_operand): Do
25001 not allow self calls to be local if the function is replaceable.
25003 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
25006 * config/rs6000/altivec.h (vec_ctz and others): Change the
25007 preprocessor macro that controls conditional compilation from
25008 _ARCH_PWR9 to __POWER9_VECTOR__.
25009 (vec_all_ne): Change parameterization of __altivec_scalar_pred
25010 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
25011 control (instead of _ARCH_PWR9 control) so that template
25012 definition uses power9-specific function.
25013 (vec_any_eq): Likewise.
25014 (vec_all_ne): Change macro definition to use a power9-specific
25015 expansion under #ifdef __POWER9_VECTOR__ control (instead of
25016 _ARCH_PWR9 control).
25017 (vec_any_eq) Likewise.
25018 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
25019 expansion for CMPNEF to remove support for xvcmpnesp instruction.
25020 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
25021 support for xvcmpnedp instruction.
25022 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
25023 macro expansion so that Power9 implementation of vec_all_ne does
25024 not use the AltiVec predicate framework.
25025 (VCMPNEH_P): Likewise.
25026 (VCMPNEW_P): Likewise.
25027 (VCMPNED_P): Likewise.
25028 (VCMPNEFP_P): Likewise.
25029 (VCMPNEDP_P): Likewise.
25030 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
25031 implementation of vec_any_eq to not use AltiVec predicate
25033 (VCMPAEH_P): Likewise.
25034 (VCMPAEW_P): Likewise.
25035 (VCMPAED_P): Likewise.
25036 (VCMPAEFP_P): Likewise.
25037 (VCMPAEDP_P): Likewise.
25038 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
25039 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
25040 not use the AltiVec predicate framework.
25041 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
25042 of vec_any_eq to not use AltiVec predicate framework.
25043 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
25044 support for predefined __POWER9_VECTOR__ macro to indicate that
25045 Power9 instruction selection is enabled.
25046 (altivec_overloaded_builtins): Remove extraneous
25047 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
25048 function argument types RS6000_BTI_bool_V16QI and
25049 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
25050 entry for overloaded function argument types RS6000_BTI_bool_V4SI
25051 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
25052 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
25053 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
25054 Power9 for implementations of vec_cmpne. Change the signature for
25055 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
25056 (representing vec_all_ne) to remove the previously described first
25057 argument of type RS6000_BTI_INTSI, as this was an artifact of
25058 reliance on the AltiVec predicate framework, which is no longer
25059 used in the implementation of these functions. Add
25060 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
25061 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
25062 since, unlike the AltiVec predicate framework implementation, we
25063 do not share function descriptors between vec_alle and vec_anyeq.
25064 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
25065 set of modes that receive special treatment even when
25066 TARGET_P9_VECTOR is true. The special treatment emits code that
25067 does not depend on Power9 instructions.
25068 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
25069 define_expand to not rely on AltiVec predicate framework.
25070 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
25072 (vector_ne_v2di_p): Change this define_expand to not rely on
25073 AltiVec predicate framework.
25074 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
25076 (vector_ne_<mode>_p): Change this define_expand to not rely on
25077 AltiVec predicate framework.
25078 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
25080 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
25081 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
25082 define_insn pattern.
25083 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
25084 define_insn pattern because the xvcmpne<VSs>. instruction is not
25086 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
25087 instruction is not supported.
25089 2017-03-01 Jakub Jelinek <jakub@redhat.com>
25091 * config/nvptx/nvptx.c: Include intl.h.
25093 2017-03-01 Martin Jambor <mjambor@suse.cz>
25096 * ipa-prop.h (ipa_bits): Removed field known.
25097 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
25098 to pointers. Adjusted their comments to warn about their sharing.
25099 (ipcp_transformation_summary): Change bits to a vector of pointers.
25100 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
25101 (ipa_get_ipa_bits_for_value): Declare.
25102 * tree-vrp.h (value_range): Mark as GTY((for_user)).
25103 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
25104 (ipa_bits_hash_table): Likewise.
25105 (ipa_vr_ggc_hash_traits): Likewise.
25106 (ipa_vr_hash_table): Likewise.
25107 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
25108 being pointers and vr_known being removed.
25109 (ipa_set_jf_unknown): Likewise.
25110 (ipa_get_ipa_bits_for_value): New function.
25111 (ipa_set_jfunc_bits): Likewise.
25112 (ipa_get_value_range): New overloaded functions.
25113 (ipa_set_jfunc_vr): Likewise.
25114 (ipa_compute_jump_functions_for_edge): Use the above functions to
25115 construct bits and vr parts of jump functions.
25116 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
25117 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
25119 (ipcp_grow_transformations_if_necessary): Also allocate
25120 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
25122 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
25123 them. Fix too long lines.
25124 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
25125 vr_known being removed.
25126 (ipa_read_jump_function): Use new setter functions to construct bits
25127 and vr parts of jump functions or set them to NULL.
25128 (write_ipcp_transformation_info): Adjust for bits being pointers.
25129 (read_ipcp_transformation_info): Likewise.
25130 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
25132 Include gt-ipa-prop.h.
25133 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
25135 (ipcp_store_bits_results): Likewise.
25136 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
25137 Do not write to existing jump functions but use a temporary instead.
25139 2017-03-01 Jakub Jelinek <jakub@redhat.com>
25142 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
25143 attempt to use its first operand as BIT_FIELD_REF base.
25145 2017-03-01 Richard Biener <rguenther@suse.de>
25147 PR middle-end/79721
25148 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
25149 interpolating formula in wrapping arithmetic.
25150 (chrec_apply): Convert chrec_evaluate return value to wanted type.
25152 2017-03-01 Jakub Jelinek <jakub@redhat.com>
25154 PR tree-optimization/79734
25155 * tree-vect-generic.c (expand_vector_condition): Optimize
25156 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
25157 Handle VEC_COND_EXPR where comparison has different inner width from
25158 type's inner width.
25160 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
25162 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
25163 markup, and similar issues. Remove @opindex entries for things
25164 that aren't options. Add missing -mmpy-option entries.
25166 2017-02-28 Jakub Jelinek <jakub@redhat.com>
25168 PR tree-optimization/79737
25169 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
25170 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
25171 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
25172 instead of byte_size. Formatting fix.
25173 (shift_bytes_in_array_right): Formatting fix.
25175 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
25178 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
25179 condition on optimize for the leaf function test.
25181 2017-02-28 Martin Liska <mliska@suse.cz>
25184 * read-rtl-function.c (function_reader::handle_unknown_directive):
25185 Bail out when one uses -flto.
25187 2017-02-28 Martin Liska <mliska@suse.cz>
25189 * common.opt: Replace space with tabular for options of <number>
25191 * config/i386/i386.opt: Show <number> value for
25192 -mlarge-data-threshold.
25193 * opts.c (print_filtered_help): Do not display number in hexadecimal
25196 2017-02-28 Martin Liska <mliska@suse.cz>
25198 * common.opt: Fix --help=option -Q for options which are of
25201 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
25203 * config/i386/i386.c (print_reg): Error out for values
25204 of 8-bit size in invalid integer register.
25206 2017-02-28 Martin Sebor <msebor@redhat.com>
25208 PR tree-optimization/79691
25209 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
25211 2017-02-28 Jakub Jelinek <jakub@redhat.com>
25214 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
25215 gcc_unreachable with output_operand_lossage.
25217 2017-02-28 Richard Biener <rguenther@suse.de>
25219 PR tree-optimization/79740
25220 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
25222 (visit_nary_op): Insert the nary into the hashtable if we
25223 pattern-matched sth.
25224 * tree-ssa-pre.c (eliminate_insert): Robustify.
25226 2017-02-28 Richard Biener <rguenther@suse.de>
25228 PR middle-end/79731
25229 * fold-const.c (decode_field_reference): Reject out-of-bound
25232 2017-02-28 Jakub Jelinek <jakub@redhat.com>
25234 * config/i386/i386.c: Include intl.h.
25235 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
25236 instead of just cond ? "..." : "...".
25237 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
25238 * coverage.c (read_counts_file): Likewise.
25239 * omp-offload.c: Include intl.h.
25240 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
25241 of just cond ? "..." : "...".
25242 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
25243 of just cond ? "..." : "...".
25245 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
25248 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
25250 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
25251 'tune for' CPU name.
25252 * config/arm/arm-cpu-data.h: Regenerated.
25254 2017-02-28 Richard Biener <rguenther@suse.de>
25256 PR tree-optimization/79732
25257 * tree-inline.c (expand_call_inline): Do not shadow var.
25259 2017-02-28 Richard Biener <rguenther@suse.de>
25261 PR tree-optimization/79723
25262 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
25263 address-space properly.
25265 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
25267 * doc/optinfo.texi (Optimization groups): Fix option used for
25269 * doc/invoke.texi (-fopt-info): Document "omp".
25270 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
25271 (OPTGROUP_ALL): Add OPTGROUP_OMP.
25272 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
25273 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
25274 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
25276 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
25278 * dumpfile.c (optgroup_options): Instead of "openmp", associate
25279 OPTGROUP_OMP with "omp".
25281 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
25284 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
25285 for arithmetic shift of unsigned V2DI.
25287 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
25289 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
25290 arc/linux.h headers.
25291 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
25292 (LINK_SPEC): Likewise.
25293 (ARC_TLS_EXTRA_START_SPEC): Likewise.
25294 (EXTRA_SPECS): Likewise.
25295 (STARTFILE_SPEC): Likewise.
25296 (ENDFILE_SPEC): Likewise.
25297 (LIB_SPEC): Likewise.
25298 (TARGET_SDATA_DEFAULT): Likewise.
25299 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
25300 (MULTILIB_DEFAULTS): Likewise.
25301 (DWARF2_UNWIND_INFO): Likewise.
25302 * config/arc/big.h: New file.
25303 * config/arc/elf.h: Likewise.
25304 * config/arc/linux.h: Likewise.
25305 * config/arc/t-uClibc: Remove.
25307 2017-02-27 Bin Cheng <bin.cheng@arm.com>
25309 PR tree-optimization/77536
25310 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
25311 (tree_transform_and_unroll_loop): Use above function to compute the
25312 estimated niter of unrolled loop and use it when scaling profile.
25313 Also use count info rather than frequency if it's non-zero.
25314 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
25315 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
25316 (vect_transform_loop): Call above function.
25318 2017-02-27 Richard Biener <rguenther@suse.de>
25320 PR tree-optimization/45397
25321 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
25322 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
25323 (visit_nary_op): Add pattern matching for CSEing sign-changed
25324 or truncated operations with wider ones.
25326 2017-02-27 Richard Biener <rguenther@suse.de>
25328 PR tree-optimization/79690
25329 * tree-vect-stmts.c (vectorizable_store): Use vector type
25330 built from the DR with address-space.
25332 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
25334 * doc/invoke.texi (Optimize Options): Refine the description
25335 of asan-use-after-return.
25337 2017-02-25 Alan Modra <amodra@gmail.com>
25339 PR rtl-optimization/79584
25340 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
25341 base, not ad->base_term, the reg within base. Remove assertion
25342 that ad->base == ad->base_term. Replace gen_int_mode using
25343 bogus mode with const0_rtx.
25345 2017-02-25 Jakub Jelinek <jakub@redhat.com>
25347 PR middle-end/79396
25348 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
25349 FMA_EXPR like tcc_binary or tcc_unary.
25351 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
25354 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
25356 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
25357 (output_loc_operands): Handle DW_OP_call_ref and
25358 DW_OP_GNU_variable_value.
25359 (struct variable_value_struct): New type.
25360 (struct variable_value_hasher): Likewise.
25361 (variable_value_hash): New variable.
25362 (string_types): Remove.
25363 (copy_loc_descr): New function.
25364 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
25365 (prepend_loc_descr_to_each): New function.
25366 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
25367 instead of add_loc_descr_to_each if the first argument is single
25368 location list and the second has multiple.
25369 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
25370 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
25371 when looking for variable value which doesn't have other location info.
25372 (loc_list_from_tree): Formatting fix.
25373 (gen_array_type_die): Simplify DW_AT_string_length handling.
25374 (adjust_string_types): Remove.
25375 (gen_subprogram_die): Don't call adjust_string_types nor test/set
25376 string_types. Call resolve_variable_values.
25377 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
25378 (resolve_addr_in_expr): Likewise. Add A argument.
25379 (copy_deref_exprloc): Remove deref argument. Adjust for the
25380 original expression being DW_OP_GNU_variable_value with optionally
25381 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
25382 optionally after it.
25383 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
25384 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
25385 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
25386 (variable_value_hasher::hash, variable_value_hasher::equal): New
25388 (resolve_variable_value_in_expr, resolve_variable_value,
25389 resolve_variable_values, note_variable_value_in_expr,
25390 note_variable_value): New functions.
25391 (dwarf2out_early_finish): Call note_variable_value on all toplevel
25394 2017-02-24 Jakub Jelinek <jakub@redhat.com>
25397 * opts.h (handle_generated_option): Add GENERATED_P argument.
25398 * opts-common.c (handle_option): Adjust function comment.
25399 (handle_generated_option): Add GENERATED_P argument, pass it to
25401 (control_warning_option): Pass false to handle_generated_option
25403 * opts.c (maybe_default_option): Pass true to handle_generated_option
25405 * optc-gen.awk: Likewise.
25407 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25409 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
25410 a REG, look at the REG it is a SUBREG of.
25411 (splitter for cmpeqsi_t): Ditto.
25413 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25415 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
25416 the special USEs with the pattern of the insn, not the insn itself.
25418 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
25421 * doc/invoke.texi: Document -mload-store-pairs.
25423 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25424 Sandra Loosemore <sandra@codesourcery.com>
25426 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
25427 argument isn't a CONST_INT.
25428 (nios2_alternate_compare_const): Assert op is a CONST_INT.
25429 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
25430 (nios2_validate_compare): Bypass alternate compare logic if *op2
25431 is not a CONST_INT.
25432 (ldstwm_operation_p): Return false if first_base is not a REG or
25433 if first_offset is not a CONST_INT.
25435 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25437 * config/cris/cris.md: Use correct operand in a define_peephole2.
25439 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25441 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
25443 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25445 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
25446 this_insn if it is an INSN or JUMP_INSN.
25447 (force_offsettable): Look at base, not at addr.
25448 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
25449 on things that aren't necessarily CONST_INTs.
25451 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
25453 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
25454 -mfpmath=sse is the default also for x86-32 targets with SSE2
25455 instruction set when @option{-ffast-math} is enabled
25457 2017-02-24 Jeff Law <law@redhat.com>
25459 PR rtl-optimizatoin/79286
25460 * ira.c (update_equiv_regs): Drop may_trap_p exception to
25463 2017-02-24 Richard Biener <rguenther@suse.de>
25465 PR tree-optimization/79389
25466 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
25469 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
25471 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
25472 function comment to reflect reality.
25473 (loop_exits_before_overflow): Fix typo in function description.
25475 2017-02-24 Richard Biener <rguenther@suse.de>
25477 PR tree-optimization/79389
25478 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
25479 properly that a threading opportunity exists. Detect conditional
25480 copy/constant propagation opportunities.
25482 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
25484 * config/visium/visium.md (type): Add trap.
25485 (b): New mode attribute.
25486 (*btst): Rename into...
25487 (*btst<mode>): ...this and adjust.
25488 (*cbranchsi4_btst_insn): Rename into...
25489 (*cbranch<mode>4_btst_insn): ...this and adjust.
25490 (trap): New define_insn.
25492 2017-02-23 Jakub Jelinek <jakub@redhat.com>
25494 PR tree-optimization/79389
25495 * ifcvt.c (struct noce_if_info): Add rev_cond field.
25496 (noce_reversed_cond_code): New function.
25497 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
25498 reversed_comparison_code. Formatting fix.
25499 (noce_try_store_flag): Test rev_cond != NULL in addition to
25500 reversed_comparison_code.
25501 (noce_try_store_flag_constants): Likewise.
25502 (noce_try_store_flag_mask): Likewise.
25503 (noce_try_addcc): Use rev_cond if non-NULL instead of
25504 reversed_comparison_code.
25505 (noce_try_cmove_arith): Likewise. Formatting fixes.
25506 (noce_try_minmax, noce_try_abs): Clear rev_cond.
25507 (noce_find_if_block): Initialize rev_cond.
25508 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
25509 instead of false as last argument never attempt to reverse it
25512 2017-02-23 Bin Cheng <bin.cheng@arm.com>
25514 PR tree-optimization/79663
25515 * tree-predcom.c (combine_chains): Process refs in reverse order
25516 only for ZERO length chains, and add explaining comment.
25518 2017-02-23 Jeff Law <law@redhat.com>
25520 PR tree-optimization/79578
25521 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
25522 in call to operand_equal_p.
25524 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
25527 * config/i386/cpuid.h: Fix another undefined behavior.
25529 2017-02-23 Richard Biener <rguenther@suse.de>
25531 PR tree-optimization/79683
25532 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
25533 vector types for data-refs.
25535 2017-02-23 Martin Liska <mliska@suse.cz>
25537 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
25539 2017-02-23 Jakub Jelinek <jakub@redhat.com>
25541 PR middle-end/79665
25542 * internal-fn.c (get_range_pos_neg): Moved to ...
25543 * tree.c (get_range_pos_neg): ... here. No longer static.
25544 * tree.h (get_range_pos_neg): New prototype.
25545 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
25546 are known to be in between 0 and signed maximum inclusive, try to
25547 expand both unsigned and signed divmod and use the cheaper one from
25550 2017-02-22 Jeff Law <law@redhat.com>
25552 PR tree-optimization/79578
25553 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
25554 to compare base operands.
25556 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
25559 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
25560 gpc_reg_operand instead of fpr_reg_operand.
25562 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
25564 * config/mips/mips.c (mips_return_in_memory): Force FP
25565 vector types to be returned in memory for o32 ABI.
25567 2017-02-22 Jakub Jelinek <jakub@redhat.com>
25569 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
25570 instead of DW_TAG_member for static data member declarations and don't
25571 set no_linkage_name for static inline data members.
25572 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
25575 2017-02-22 Martin Liska <mliska@suse.cz>
25577 * doc/invoke.texi: Replace inequality signs with square brackets
25580 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25582 PR tree-optimization/68644
25583 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
25585 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
25588 * lra-constraints.c (simplify_operand_subreg): Handle
25589 WORD_REGISTER_OPERATIONS targets.
25591 2017-02-22 Jakub Jelinek <jakub@redhat.com>
25594 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
25595 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
25596 elimination by swapping fld*.
25598 2017-02-22 Richard Biener <rguenther@suse.de>
25600 PR tree-optimization/79673
25601 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
25602 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
25603 irrelevant address-space qualifiers and avoiding a
25604 ADDR_SPACE_CONVERT_EXPR from fold_convert.
25606 2017-02-22 Richard Biener <rguenther@suse.de>
25608 PR tree-optimization/79666
25609 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
25610 to not symbolically negate if that may introduce undefined
25613 2017-02-22 Martin Liska <mliska@suse.cz>
25616 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
25617 * data-streamer-out.c (streamer_write_gcov_count_stream):
25619 * value-prof.c (stream_out_histogram_value): Make assert more
25620 precise based on type of counter.
25622 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
25625 * config/i386/i386.md (standard_x87sse_constant_load splitter):
25626 Use nonimmediate_operand instead of memory_operand for operand 1.
25627 (float-extend standard_x87sse_constant_load splitter): Ditto.
25629 2017-02-21 Jeff Law <law@redhat.com>
25631 PR tree-optimization/79621
25632 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
25633 blocks with edges to themselves.
25635 2017-02-21 Jakub Jelinek <jakub@redhat.com>
25638 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
25639 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
25640 Use gimple_call_builtin_p.
25643 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
25644 on temporarily removed DEBUG_INSNs.
25646 PR tree-optimization/79649
25647 * tree-loop-distribution.c (classify_partition): Give up on
25648 non-generic address space loads/stores.
25650 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
25652 * doc/loop.texi (Loop manipulation): Remove nonexistent
25653 tree_ssa_loop_version from the documentation.
25654 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
25656 2017-02-21 Jakub Jelinek <jakub@redhat.com>
25659 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
25660 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
25661 * config/rs6000/rs6000.c: Include except.h.
25662 (rs6000_expand_split_stack_prologue): Call
25663 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
25665 2017-02-21 Martin Jambor <mjambor@suse.cz>
25668 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
25669 have been analyzed.
25671 2017-02-21 Martin Jambor <mjambor@suse.cz>
25673 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
25674 for backward compatibility only.
25675 * doc/invoke.texi (Option Summary): Remove all references to
25676 -fipa-cp-alignment.
25678 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
25682 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25684 * lra-constraints.c (curr_insn_transform): Handle
25685 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
25687 2017-02-21 Martin Liska <mliska@suse.cz>
25689 * config/i386/i386.opt: Replace -masm-dialect with -masm.
25691 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
25693 PR translation/79638
25694 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
25696 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
25699 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
25700 (arm_function_ok_for_sibcall): Return false for an indirect call by
25701 descriptor if all the argument registers are used.
25702 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
25703 alignment of the function.
25705 2017-02-21 Jakub Jelinek <jakub@redhat.com>
25707 PR tree-optimization/61441
25708 * simplify-rtx.c (simplify_const_unary_operation): For
25709 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
25710 the sNaN unmodified.
25712 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
25714 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25715 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25716 instead of SYSTEM_HEADER_DIR.
25718 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
25719 Martin Liška <mliska@suse.cz>
25721 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
25722 Fix typos and grammar, use active voice, and clarify.
25724 2017-02-20 Marek Polacek <polacek@redhat.com>
25726 PR middle-end/79537
25727 * gimplify.c (gimplify_expr): Handle unused *&&L;.
25730 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
25732 2017-02-20 Jakub Jelinek <jakub@redhat.com>
25735 * config/i386/i386.c (ix86_expand_builtin): Handle
25736 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
25737 ix86_builtins_isa[fcode].isa as a requirement of those
25738 flags and any other flag in the bitmask.
25739 (ix86_init_mmx_sse_builtins): Use 0 instead of
25740 ~OPTION_MASK_ISA_64BIT as mask.
25741 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
25742 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
25743 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
25744 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
25746 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25749 * lra-constraints.c (split_reg): Check requested split mode
25750 is supported by the register.
25752 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25754 * lra-constraints.c (simplify_operand_subreg): Remove early
25757 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25760 * lra-constraints.c (curr_insn_transform): Tighten condition
25761 for converting SUBREG reloads from OP_OUT to OP_INOUT.
25763 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25766 * lra-constraints.c (curr_insn_transform): Handle
25767 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
25769 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
25772 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
25774 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
25776 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
25779 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
25782 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25784 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
25787 2017-02-18 Jakub Jelinek <jakub@redhat.com>
25789 * final.c (last_columnnum, override_columnnum): New variables.
25790 (final_start_function): Set last_columnnum, pass it to begin_prologue
25791 hook and pass 0 to dwarf2out_begin_prologue.
25792 (final_scan_insn): Update override_columnnum. Pass last_columnnum
25793 to source_line debug hook.
25794 (notice_source_line): Compute last_columnnum and for debug_column_info
25795 return true on column changes.
25796 * debug.h (struct gcc_debug_hooks): Add column argument to
25797 source_line and begin_prologue hooks.
25798 (debug_nothing_int_charstar_int_bool): Remove prototype.
25799 (debug_nothing_int_int_charstar,
25800 debug_nothing_int_int_charstar_int_bool): New prototypes.
25801 (dwarf2out_begin_prologue): Add column argument.
25802 * debug.c (do_nothing_debug_hooks): Adjust source_line and
25803 begin_prologue hooks.
25804 (debug_nothing_int_charstar_int_bool): Remove.
25805 (debug_nothing_int_int_charstar,
25806 debug_nothing_int_int_charstar_int_bool): New functions.
25807 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
25808 through to dwarf2out_source_line.
25809 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
25810 (dwarf2out_source_line): Add column argument, emit it if requested.
25811 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
25813 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
25814 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
25815 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
25816 through to dwarf2out_begin_prologue.
25817 (vmsdbgout_source_line): Add column argument, pass it through to
25818 dwarf2out_source_line.
25819 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
25820 dbxout_source_line caller.
25821 (dbxout_source_line): Add column argument.
25823 * common.opt (gno-column-info, gcolumn-info): New options.
25824 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
25825 (check_die): Also test for multiple DW_AT_decl_column attributes.
25826 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
25827 DW_AT_decl_column if requested.
25828 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
25830 (gen_variable_die): Likewise.
25831 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
25832 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
25835 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
25836 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
25837 (ix86_handle_option): Handle OPT_m3dnowa.
25838 * doc/invoke.texi (-m3dnowa): Document.
25839 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
25840 -m3dnowa instead of -m3dnow -march=athlon.
25843 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
25844 instead of gcc_assert for K, r and R code checks. Formatting fixes.
25846 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25849 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
25850 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
25851 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
25852 generator for vsx_xxpermdi_<mode>_be.
25853 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
25854 force big-endian semantics.
25855 (vsx_xxpermdi_<mode>_be): New define_expand with same
25856 implementation as previous version of vsx_xxpermdi_<mode>.
25858 2017-02-17 Jakub Jelinek <jakub@redhat.com>
25860 PR tree-optimization/79327
25861 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
25862 variable, its initialization and use.
25864 2017-02-17 Julia Koval <julia.koval@intel.com>
25866 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
25867 (OPTION_MASK_ISA_PKU_UNSET): New.
25868 (ix86_handle_option): Handle -mrdpid.
25869 * config/i386/cpuid.h (bit_RDPID): New.
25870 * config/i386/driver-i386.c (host_detect_local_cpu):
25871 Detect RDPID feature.
25872 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
25873 * config/i386/i386-c.c (ix86_target_macros_internal):
25875 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
25876 (ix86_valid_target_attribute_inner_p): Add "rdpid".
25877 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
25878 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
25879 * config/i386/i386.md (define_insn "rdpid"): New.
25880 * config/i386/i386.opt Add -mrdpid.
25881 * config/i386/immintrin.h (_rdpid_u32): New.
25883 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
25885 PR rtl-optimization/79541
25886 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
25887 instead of transforming it into USE.
25889 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
25891 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
25892 If HONOR_SNANS (SFmode) force the input to a register.
25893 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
25894 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
25895 an frsp or similar insn.
25897 2017-02-17 Martin Liska <mliska@suse.cz>
25899 PR rtl-optimization/79577
25900 * params.def (selsched-max-sched-times): Increase minimum to 1.
25902 2017-02-17 Martin Liska <mliska@suse.cz>
25904 PR rtl-optimization/79574
25905 * gcse.c (want_to_gcse_p): Prevent integer overflow.
25907 2017-02-17 Martin Liska <mliska@suse.cz>
25909 PR tree-optimization/79529
25910 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
25911 ssa_defined_default_def_p to handle cases which are implicitly
25913 * tree-ssa.c (ssa_defined_default_def_p): New function.
25914 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
25915 which are implicitly defined.
25916 * tree-ssa.h (ssa_defined_default_def_p): Declare.
25918 2017-02-17 Richard Biener <rguenther@suse.de>
25920 PR middle-end/79576
25921 * params.def (max-ssa-name-query-depth): Limit to 10.
25923 2017-02-17 Richard Biener <rguenther@suse.de>
25925 PR tree-optimization/79552
25926 * tree-ssa-structalias.c (visit_loadstore): Properly verify
25929 2017-02-17 Richard Biener <rguenther@suse.de>
25932 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
25934 2017-02-17 Marek Polacek <polacek@redhat.com>
25936 PR middle-end/79536
25937 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
25938 (fold_negate_expr): New wrapper.
25940 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
25942 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
25943 Correct terminology and de-emphasize pre-standard behavior.
25945 2017-02-16 Alan Modra <amodra@gmail.com>
25947 PR rtl-optimization/79286
25948 * ira.c (def_dominates_uses): New function.
25949 (update_equiv_regs): Don't create an equivalence for insns that
25950 may trap where the register def does not dominate the use.
25952 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
25954 PR rtl-optimization/78127
25955 * lra.c (lra): Call lra_eliminate before finish the loop after
25958 2017-02-16 Richard Biener <rguenther@suse.de>
25960 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
25962 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
25963 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
25964 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
25965 (isl_val_int_from_wi): New function.
25966 (extract_affine_gmp): Rename to ...
25967 (extract_affine_wi): ... this, take a widest_int.
25968 (extract_affine_int): Just wrap extract_affine_wi.
25969 (add_param_constraints): Use isl_val_int_from_wi.
25970 (add_loop_constraints): Likewise, and extract_affine_wi.
25972 2017-02-15 Jeff Law <law@redhat.com>
25974 PR middle-end/79521
25975 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
25976 ira_init_register_move_cost_if_necessary.
25978 2017-02-15 Martin Sebor <msebor@redhat.com>
25980 PR middle-end/32003
25981 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
25982 removed in a prior commit.
25984 2017-02-15 Bin Cheng <bin.cheng@arm.com>
25986 PR tree-optimization/79347
25987 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
25988 counters during peeling.
25990 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
25992 * Makefile.in (site.exp): Remove "set ISLVER".
25994 2017-02-15 Jakub Jelinek <jakub@redhat.com>
25997 * real.c (real_from_integer): Call real_convert even for decimal.
25999 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
26002 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
26004 2017-02-14 Andrew Pinski <apinski@cavium.com>
26006 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
26007 cores and change the partno/implementer to be correct.
26008 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
26009 the 'B" as the implementer.
26010 * config/aarch64/aarch64-tune.md: Regenerate.
26012 2017-02-14 Carl Love <cel@us.ibm.com>
26014 * config/rs6000/rs6000.c: Add case statement entry to make the
26015 xvcvuxdsp built-in argument unsigned.
26016 * config/rs6000/vsx.md: Fix the source and return operand types so they
26017 match the instruction definitions from the ISA document. Fix typo
26018 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
26021 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
26024 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
26025 member early_clobber_alts.
26026 * lra-lives.c (reg_early_clobber_p): New.
26027 (process_bb_lives): Use it.
26028 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
26029 (debug_operand_data): Initialize early_clobber_alts.
26030 (setup_operand_alternative): Set up early_clobber_alts.
26031 (collect_non_operand_hard_regs): Ditto. Pass early clobber
26032 alternatives to new_insn_reg.
26033 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
26035 (lra_update_insn_regno_info): Pass the new arg.
26037 2017-02-14 Jakub Jelinek <jakub@redhat.com>
26039 PR middle-end/79505
26040 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
26041 (new_oacc_loop_raw): Don't clear already cleared fields.
26044 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
26045 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
26046 _mm512_prefetch_i64gather_ps): New inline functions and macros.
26048 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
26051 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
26053 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
26056 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
26057 the extra instruction to the right place to store 128-bit constant
26060 2017-02-14 Martin Sebor <msebor@redhat.com>
26062 PR middle-end/79448
26063 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
26064 warning for strings of unknown length.
26066 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
26068 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
26070 2017-02-14 Jeff Law <law@redhat.com>
26073 * ira-costs.c (scan_one_insn): Initialize register move costs
26074 for pseudos seen in USE/CLOBBER insns.
26076 PR tree-optimization/79095
26077 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
26078 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
26079 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
26080 if the operands are known to be not equal, then the resulting range
26082 (intersect_ranges): If the new range is ~[0,0] and the old range is
26083 wide, then prefer ~[0,0].
26084 * tree-vrp.c (overflow_comparison_p_1): New function.
26085 (overflow_comparison_p): New function.
26086 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
26087 if NAME is used in an overflow test.
26088 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
26089 overflow check that can be expressed as an equality test, then adjust
26090 ops to be that equality test.
26092 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26094 * config/s390/s390-builtin-types.def: Remove flags argument.
26095 * config/s390/s390.c (s390_init_builtins): Likewise.
26097 2017-02-14 Martin Liska <mliska@suse.cz>
26099 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
26100 vector. Fix trailing white spaces.
26102 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
26104 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
26107 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26109 PR rtl-optimization/68664
26110 * config/arm/arm.c (arm_sched_can_speculate_insn):
26111 New function. Declare prototype.
26112 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
26114 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26116 PR rtl-optimization/68664
26117 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
26119 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
26121 2017-02-14 Amit Pawar <amit.pawar@amd.com>
26123 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
26124 max skip bytes for function, loop and jump.
26126 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26128 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
26129 ABS_EXPR for gimple dump.
26131 2017-02-14 Jakub Jelinek <jakub@redhat.com>
26134 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
26136 PR tree-optimization/79408
26137 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
26138 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
26139 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
26140 also if rhs1 is INTEGER_CST.
26142 2017-02-14 Richard Biener <rguenther@suse.de>
26144 PR middle-end/79432
26145 * tree-into-ssa.c (insert_phi_nodes): When the function can
26146 have abnormal edges rewrite SSA names with broken use-def
26147 dominance out of SSA and register them for PHI insertion.
26149 2017-02-13 Martin Sebor <msebor@redhat.com>
26151 PR middle-end/79496
26152 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
26153 clearing info.nowrite flag when snprintf size argument is a range.
26155 2017-02-13 Jakub Jelinek <jakub@redhat.com>
26157 * cprop.c (cprop_jump): Add missing space in string literal.
26158 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
26159 (get_constraint_for_component_ref): Likewise.
26160 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
26161 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
26162 * lra-constraints.c (process_alt_operands): Likewise.
26163 * ipa-inline.c (inline_small_functions): Likewise.
26164 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
26165 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
26166 * trans-mem.c (diagnose_tm_1_op): Likewise.
26167 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
26168 (grid_parallel_clauses_gridifiable): Likewise.
26170 * config/nvptx/mkoffload.c (process): Add space in between
26173 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
26174 "MOD4_SSE_REGS" and "ALL_REGS".
26176 * spellcheck.c (test_data): Add , in between "foo" and "food".
26178 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26181 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
26182 boundary crossing check and subsequent code generation agree.
26184 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26186 * config/aarch64/aarch64.c (has_memory_op): Delete.
26187 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
26190 2017-02-13 Jakub Jelinek <jakub@redhat.com>
26192 PR rtl-optimization/79388
26193 PR rtl-optimization/79450
26194 * combine.c (distribute_notes): When removing TEM_INSN for which
26195 corresponding dest has last value recorded, invalidate that last
26198 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26200 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
26201 of explicit '@'. Add missing assembly comment marker on branch costs
26204 2017-02-13 Nathan Sidwell <nathan@acm.org>
26206 * gengtype-lex.l (<in_struct>): Add '/'.
26208 2017-02-13 Martin Liska <mliska@suse.cz>
26211 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
26213 2017-02-13 Richard Biener <rguenther@suse.de>
26215 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
26217 * configure: Re-generate.
26218 * config.in: Likewise.
26219 * graphite-dependences.c: Simplify as if
26220 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
26221 * graphite-isl-ast-to-gimple.c: Likewise.
26222 * graphite-optimize-isl.c: Likewise.
26223 * graphite-poly.c: Likewise.
26224 * graphite-sese-to-poly.c: Likewise.
26225 * graphite.h: Likewise.
26226 * toplev.c: Include isl/version.h and use isl_version () for
26227 printing the ISL version.
26228 * doc/install.texi: Update ISL requirement.
26230 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
26232 * doc/standards.texi (Standards): Update reference to
26235 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
26237 * doc/extend.texi (Named Address Spaces): sourceware.org now
26239 * doc/install.texi (Binaries): Ditto.
26242 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
26244 * doc/cpp.texi: Replace "stringify"/"stringification" with C
26245 standard terminology "stringize"/"stringizing" throughout.
26246 * doc/cppinternals.texi: Likewise.
26248 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
26250 * doc/extend.texi: Fix some spelling mistakes and typos.
26251 * doc/invoke.texi: Likewise.
26253 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
26256 * params.def (inline-min-speedup) Change from 10 to 8.
26258 2017-02-11 Jakub Jelinek <jakub@redhat.com>
26260 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
26263 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
26266 * ipa-inline-analysis.c (get_minimal_bb): New function.
26267 (record_modified): Use it.
26268 (remap_edge_change_prob): Handle also ancestor functions.
26270 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
26272 * doc/contrib.texi (Contributors): Remove broken link into
26273 the Mauve CVS repository.
26275 2017-02-11 Jakub Jelinek <jakub@redhat.com>
26277 PR middle-end/79454
26278 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
26279 result computation whenever lhs doesn't have vector mode, not
26280 just when it has BLKmode.
26282 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
26284 * doc/makefile.texi (profiledbootstrap): Refer to the
26285 installation instructions only in textual form.
26287 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26290 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
26292 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
26294 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
26295 (Specific): Update mingw-w64 reference.
26297 (Specific): Remove broken link to Renesas RX processor.
26299 2017-02-10 Richard Biener <rguenther@suse.de>
26301 * toplev.c (process_options): Do not mention obsolete graphite
26302 options when printing sorry message about missing graphite support.
26303 Mention -floop-nest-optimize.
26305 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
26307 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
26308 (vtst_p16): Likewise.
26309 (vtstq_p8): Likewise.
26310 (vtstq_p16): Likewise.
26312 (vtstq_p64): Likewise.
26313 * config/arm/arm_neon.h (vgetq_lane_p64): New.
26314 (vset_lane_p64): New.
26315 (vsetq_lane_p64): New.
26317 2017-02-10 Jakub Jelinek <jakub@redhat.com>
26319 PR tree-optimization/79411
26320 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
26321 stmt operands are SSA_NAMEs used in abnormal phis.
26322 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
26325 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
26328 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
26331 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
26333 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
26335 2017-02-09 Jakub Jelinek <jakub@redhat.com>
26337 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
26341 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
26342 not arbitrary TREE_CONSTANT.
26345 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
26346 "omp declare target link" attribute unless is_global_var.
26347 * omp-offload.c (find_link_var_op): Likewise.
26349 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
26350 Chung-Lin Tang <cltang@codesourcery.com>
26352 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
26354 (gimplify_adjust_omp_clauses): Don't delete TILE.
26355 (gimplify_omp_for): Deal with TILE.
26356 * internal-fn.c (expand_GOACC_TILE): New function.
26357 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
26359 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
26360 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
26362 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
26363 avoid DIV for outermost collapse var.
26364 (expand_oacc_for): Insert tile element loop as needed. Adjust.
26365 Remove out of date comments, fix whitespace.
26366 * omp-general.c (omp_extract_for_data): Deal with tiling.
26367 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
26368 adjust OLF_DIM_BASE value.
26369 (struct omp_for_data): Add tiling field.
26370 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
26371 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
26372 for auto loops. Remove default auto determining, moved to
26373 oacc_loop_fixed_partitions.
26374 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
26375 stmts, add e_mask field.
26376 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
26377 (oacc_thread_numbers): Use oacc_dim_call.
26378 (oacc_xform_tile): New.
26379 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
26380 (finish_oacc_loop): Adjust for ifns vector.
26381 (oacc_loop_discover_walk): Append loop abstraction sites to list,
26382 add case for GOACC_TILE fns.
26383 (oacc_loop_xform_loop): Delete.
26384 (oacc_loop_process): Iterate over call list directly, and add
26385 handling for GOACC_TILE fns.
26386 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
26388 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
26389 vector partitioning to outer loops. Assign 2 partitions to loops
26390 when available. Add TILE handling.
26391 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
26392 (execite_oacc_device_lower): Process GOACC_TILE fns,
26393 ignore unknown specs.
26394 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
26395 * tree.c (omp_clause_num_ops): Adjust TILE ops.
26396 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
26398 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
26400 * configure.ac (ACX_BUGURL): Update.
26401 * configure: Regenerate.
26403 2017-02-09 Richard Biener <rguenther@suse.de>
26405 PR tree-optimization/69823
26406 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
26407 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
26409 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
26411 * config/arc/arc-c.def: Add __NPS400__ definition.
26412 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
26413 (TARGET_NPS400): Define.
26415 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
26417 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
26419 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
26420 pointer, arch_info.
26421 (arc_cpu_types): Fill the arch_info field with a pointer into the
26422 arc_arch_types table.
26423 (arc_selected_cpu): Declare.
26424 * config/arc/arc.c (arc_selected_cpu): Make global.
26425 (arc_selected_arch): Delete.
26426 (arc_base_cpu): Delete.
26427 (arc_override_options): Remove references to deleted variables,
26428 update access to arch information.
26429 (ARC_OPT): Update access to arch information.
26430 (ARC_OPTX): Likewise.
26431 * config/arc/arc.h (arc_base_cpu): Remove declaration.
26432 (TARGET_ARC600): Update access to arch information.
26433 (TARGET_ARC601): Likewise.
26434 (TARGET_ARC700): Likewise.
26435 (TARGET_EM): Likewise.
26436 (TARGET_HS): Likewise.
26437 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
26440 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
26443 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
26444 condition/operands for integer GE/LE/GEU/LEU operations.
26446 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
26448 PR translation/79397
26449 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
26452 2017-02-08 Martin Jambor <mjambor@suse.cz>
26455 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
26456 whether allocation happened.
26457 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
26458 nothing was allocated.
26460 2017-02-08 Jakub Jelinek <jakub@redhat.com>
26462 PR tree-optimization/79408
26463 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
26464 constant, but SSA_NAME with a known integer range, use the minimum
26465 of that range instead of op1 to determine if modulo can be replaced
26466 with its first operand.
26468 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26470 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
26472 2017-02-08 Richard Biener <rguenther@suse.de>
26474 PR tree-optimization/71824
26475 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
26476 Check all loops contained in the merged region.
26478 2017-02-07 Andrew Pinski <apinski@cavium.com>
26480 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
26482 2017-02-07 Andrew Pinski <apinski@cavium.com>
26484 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
26485 (thunderxt88): Likewise.
26486 (thunderxt81): Disable LSE and change v8.1 to v8.
26487 (thunderxt83): Likewise.
26489 2017-02-07 Jakub Jelinek <jakub@redhat.com>
26490 Richard Biener <rguenther@suse.de>
26492 PR middle-end/79399
26493 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
26494 type from int to size_t.
26495 * ira-costs.c (struct_costs_size): Change type from int to size_t.
26497 2017-02-07 Jakub Jelinek <jakub@redhat.com>
26499 PR rtl-optimization/79386
26500 * cprop.c (bypass_conditional_jumps): Initialize
26501 bypass_last_basic_block already before splitting bbs after
26502 unconditional traps...
26503 (bypass_conditional_jumps): ... rather than here.
26506 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
26507 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
26508 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
26509 fix -masm=intel patterns.
26511 2017-02-07 Richard Biener <rguenther@suse.de>
26513 PR tree-optimization/79256
26514 PR middle-end/79278
26515 * builtins.c (get_object_alignment_2): Use min_align_of_type
26516 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
26517 and ADJUST_FIELD_ALIGN.
26519 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
26521 * doc/tm.texi: Regenerate.
26522 * stor-layout.c (layout_decl): Adjust.
26523 (update_alignment_for_field): Likewise.
26524 (place_field): Likewise.
26525 (min_align_of_type): Likewise.
26526 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
26527 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
26528 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
26529 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
26530 * config/frv/frv.c (frv_adjust_field_align): Likewise.
26531 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
26532 * config/i386/i386.c (x86_field_alignment): Likewise.
26533 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
26534 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
26535 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
26536 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
26537 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
26538 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
26542 2017-01-30 Richard Biener <rguenther@suse.de>
26544 PR tree-optimization/79256
26545 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
26546 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
26549 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
26551 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
26552 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
26553 builtins to SImode and emit a zero-extend, if necessary.
26555 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
26557 * docs/invoke.texi (RISC-V Options): Alphabetize.
26559 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
26561 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
26564 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
26566 * config/riscv/riscv.c: New file.
26567 * common/config/riscv/riscv-common.c: Likewise.
26568 * config.gcc: Likewise.
26569 * config/riscv/constraints.md: Likewise.
26570 * config/riscv/elf.h: Likewise.
26571 * config/riscv/generic.md: Likewise.
26572 * config/riscv/linux.h: Likewise.
26573 * config/riscv/multilib-generator: Likewise.
26574 * config/riscv/peephole.md: Likewise.
26575 * config/riscv/pic.md: Likewise.
26576 * config/riscv/predicates.md: Likewise.
26577 * config/riscv/riscv-builtins.c: Likewise.
26578 * config/riscv/riscv-c.c: Likewise.
26579 * config/riscv/riscv-ftypes.def: Likewise.
26580 * config/riscv/riscv-modes.def: Likewise.
26581 * config/riscv/riscv-opts.h: Likewise.
26582 * config/riscv/riscv-protos.h: Likewise.
26583 * config/riscv/riscv.h: Likewise.
26584 * config/riscv/riscv.md: Likewise.
26585 * config/riscv/riscv.opt: Likewise.
26586 * config/riscv/sync.md: Likewise.
26587 * config/riscv/t-elf-multilib: Likewise.
26588 * config/riscv/t-linux: Likewise.
26589 * config/riscv/t-linux-multilib: Likewise.
26590 * config/riscv/t-riscv: Likewise.
26591 * configure.ac: Likewise.
26592 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
26593 Waterman as RISC-V maintainers.
26594 * doc/install.texi: Add RISC-V entries.
26595 * doc/invoke.texi: Add RISC-V options section.
26596 * doc/md.texi: Add RISC-V constraints section.
26597 * configure: Regenerated.
26599 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
26602 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
26603 false values to be constant vectors with all 0 or all 1 bits set.
26604 (vcondu<mode><mode>): Likewise.
26605 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
26607 (fpmask_comparison_operator): Update comment.
26608 (vecint_comparison_operator): New predicate.
26609 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
26610 vector conditionals when the true and false values are constant
26611 vectors with all 0 bits or all 1 bits set.
26613 2017-02-06 Martin Sebor <msebor@redhat.com>
26615 PR tree-optimization/79376
26616 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
26618 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
26620 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
26621 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
26622 to simplify split condition.
26624 2017-02-06 Jakub Jelinek <jakub@redhat.com>
26626 * omp-expand.c (oxpand_omp_atomic_fetch_op,
26627 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
26630 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
26632 PR rtl-optimization/68664
26633 * target.def (can_speculate_insn): New hook.
26634 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
26635 * doc/tm.texi: Regenerate.
26636 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
26637 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
26638 (rs6000_sched_can_speculate_insn): New function.
26640 2017-02-06 Jakub Jelinek <jakub@redhat.com>
26642 PR tree-optimization/79284
26643 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
26644 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
26645 vectorizable_mask_load_store, vectorizable_operation,
26646 vect_is_simple_cond, get_same_sized_vectype): Use it instead
26647 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
26648 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
26649 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
26650 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
26651 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
26652 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
26653 is_gimple_assign (stmt). Replace another such test with
26654 is_gimple_assign (stmt).
26656 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
26659 * config/avr/avr.c (rtl-iter.h): Include it.
26660 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
26661 (avr_legitimate_combined_insn): ...and implementation.
26663 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
26665 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
26666 * config/s390/s390.c (s390_const_operand_ok)
26667 (s390_canonicalize_comparison, s390_extract_part)
26668 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
26669 (s390_contiguous_bitmask_p, s390_rtx_costs)
26670 (legitimize_pic_address): Likewise.
26671 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
26672 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
26673 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
26674 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
26675 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
26677 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
26679 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
26680 REGNO($0) == REGNO($1).
26682 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26684 * config/s390/linux.h(SIZE_TYPE): Add comment.
26686 2017-02-06 Julian Brown <julian@codesourcery.com>
26687 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26688 Virendra Pathak <virendra.pathak@broadcom.com>
26690 * config/aarch64/aarch64-cores.def: Change the scheduler
26692 * config/aarch64/aarch64.md: Include thunderx2t99.md.
26693 * config/aarch64/thunderx2t99.md: New file.
26695 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
26697 * doc/standards.texi (Go Language): Update link to language
26700 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
26702 * tree-eh.c (lower_resx): Sanitize profile.
26703 (cleanup_empty_eh_move_lp): Likewise.
26705 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
26708 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
26710 * cfgloopmanip.h (loop_version): Update prototype.
26711 * modulo-sched.c (sms_schedule): Update call of loop_version.
26712 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
26713 * tree-parloops.c (gen_parallel_loop): Likewise.
26714 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
26715 * tree-ssa-loop-split.c (split_loop): Likewise.
26716 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
26717 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
26719 2017-02-05 Martin Liska <mliska@suse.cz>
26722 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
26724 (print_operand_address): Initialize a struct to zero.
26726 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
26728 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
26729 garbage collector only in textual form.
26731 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
26733 * doc/extend.texi (x86 specific memory model extensions for
26734 transactional memory): Simplify a phrase.
26736 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
26739 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
26740 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
26741 (atomic_storedi_1): Likewise.
26743 2017-02-04 Jakub Jelinek <jakub@redhat.com>
26745 PR tree-optimization/79338
26746 * tree-parloops.c (gather_scalar_reductions): Don't call
26747 vect_analyze_loop_form for loop->inner before destroying loop's
26750 2017-02-03 Martin Sebor <msebor@redhat.com>
26752 PR tree-optimization/79327
26753 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
26754 when precision has resulted in leading zeros.
26755 (format_integer): Adjust the likely counter to assume an unknown
26756 argument that may be zero is non-zero.
26758 2017-02-03 Jason Merrill <jason@redhat.com>
26761 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
26762 avoid copying non-taken branch.
26764 2017-02-03 Jakub Jelinek <jakub@redhat.com>
26766 PR tree-optimization/79340
26767 * tree-vect-loop.c (vectorizable_reduction): Release
26768 vec_defs elements after safe_splicing them into other vectors.
26771 PR tree-optimization/79327
26772 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
26773 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
26775 (format_integer): Use wide_int_to_tree instead of build_int_cst
26776 + to_?hwi. If argmin is NULL, just set argmin and argmax to
26777 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
26778 of shortest and longest sequence.
26780 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
26782 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
26783 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
26785 2017-02-03 Walter Lee <walt@tilera.com>
26788 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
26789 after initial stackframe link reg save.
26790 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
26792 2017-02-03 Jakub Jelinek <jakub@redhat.com>
26795 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
26796 wu for stxssp alternative.
26798 2017-02-03 Martin Sebor <msebor@redhat.com>
26800 PR tree-optimization/79352
26801 * gimple-fold.c (get_range_strlen): Add argument.
26802 (get_range_strlen): Change return type to bool.
26803 (get_maxval_strlen): Pass in a dummy argument.
26804 * gimple-fold.h (get_range_strlen): Change return type to bool.
26805 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
26806 * tree.h (array_at_struct_end_p): Add argument.
26807 * tree.c (array_at_struct_end_p): Handle it.
26809 2017-02-03 Martin Liska <mliska@suse.cz>
26812 * multiple_target.c (create_dispatcher_calls): Redirect edge
26813 from a caller of a dispatcher.
26814 (expand_target_clones): Make the clones local.
26815 (ipa_target_clone): Do both target clones and resolvers.
26816 (ipa_dispatcher_calls): Remove the pass.
26817 (pass_dispatcher_calls::gate): Likewise.
26818 (make_pass_dispatcher_calls): Likewise.
26819 * passes.def (pass_target_clone): Put as very first IPA early
26822 2017-02-03 Martin Liska <mliska@suse.cz>
26824 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
26825 in case of a function with ifunc attribute.
26827 2017-02-03 Martin Liska <mliska@suse.cz>
26829 * cgraph.c (cgraph_node::dump): Dump function version info.
26830 * symtab.c (symtab_node::dump_base): Add missing new line.
26832 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
26834 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
26835 (ifcombine_ifandif): Use it.
26837 2017-02-03 Martin Liska <mliska@suse.cz>
26839 * doc/invoke.texi: Document default value for
26840 use-after-scope-direct-emission-threshold.
26842 2017-02-03 Martin Liska <mliska@suse.cz>
26844 PR tree-optimization/79339
26845 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
26846 (format_floating): Likewise.
26848 2017-02-03 Martin Liska <mliska@suse.cz>
26851 * ipa-prop.c (ipa_node_params_t::insert): Remove current
26853 (ipa_node_params_t::remove): Likewise.
26854 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
26855 initialization from removed ipa_node_params_t::insert.
26856 (ipa_node_params::~ipa_node_params): Move from removed
26857 ipa_node_params_t::release.
26858 * symbol-summary.h (symbol_summary::m_released): New member.
26859 Do not release a summary twice. Do not allow to call finalizer
26860 for types of a summary that live in GGC memory.
26862 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26864 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
26867 2017-02-02 Martin Sebor <msebor@redhat.com>
26869 PR middle-end/79275
26870 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
26871 (format_string): Tighten up the range of output for non-constant
26872 strings and correct the expected range for wide non-constant strings.
26874 2017-02-02 Martin Sebor <msebor@redhat.com>
26876 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
26878 PR middle-end/32003
26879 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
26881 (-fdump-tree-@var): Add to index and document how to come up
26882 with pass-specific option and dump file names.
26883 (-fdump-passes): Clarify where to look for output.
26885 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
26887 PR middle-end/77445
26888 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
26889 statistics of the analyzed path; allow threading for speed when
26890 any of BBs along the path are optimized for speed.
26892 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
26894 PR middle-end/78468
26895 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
26896 settings of the virtual registers.
26899 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
26901 * explow.c (get_dynamic_stack_size): Take known alignment of stack
26902 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
26905 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26907 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
26908 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
26910 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26912 * config/s390/s390.md: Add missing comments with the expanded
26914 * config/s390/vector.md: Likewise.
26915 * config/s390/vx-builtins.md: Likewise.
26917 2017-02-02 Jakub Jelinek <jakub@redhat.com>
26920 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
26921 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
26922 conditions on a single line.
26924 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26926 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
26927 __S390_VX__ to __VX__.
26929 2017-02-01 Andrew Pinski <apinski@cavium.com>
26931 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
26932 stmt_info to record_stmt_cost.
26933 (vect_get_known_peeling_cost): Pass stmt_info if known to
26935 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
26936 cpu_vector_cost field into
26937 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
26938 field into vec_int_stmt_cost and vec_fp_stmt_cost.
26939 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
26940 splitting of scalar_stmt_cost and vec_stmt_cost.
26941 (thunderx_vector_cost): Likewise.
26942 (cortexa57_vector_cost): LIkewise.
26943 (exynosm1_vector_cost): Likewise.
26944 (xgene1_vector_cost): Likewise.
26945 (thunderx2t99_vector_cost): Improve after the splitting of the two
26947 (aarch64_builtin_vectorization_cost): Update for the splitting of
26948 scalar_stmt_cost and vec_stmt_cost.
26950 2017-02-01 Torvald Riegel <triegel@redhat.com>
26951 Richard Henderson <rth@redhat.com>
26953 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
26954 conditional on existance of a fast atomic load.
26955 * optabs-query.c (can_atomic_load_p): New function.
26956 * optabs-query.h (can_atomic_load_p): Declare it.
26957 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
26958 no fast atomic load is available for the particular size of access.
26959 (expand_atomic_compare_and_swap): Likewise.
26960 (expand_atomic_load): Likewise.
26961 (expand_atomic_store): Likewise.
26962 (expand_atomic_fetch_op): Likewise.
26963 * testsuite/lib/target-supports.exp
26964 (check_effective_target_sync_int_128): Remove x86 because it provides
26965 no fast atomic load.
26966 (check_effective_target_sync_int_128_runtime): Likewise.
26968 2017-02-01 Richard Biener <rguenther@suse.de>
26970 * graphite.c: Include tree-vectorizer.h for find_loop_location.
26971 (graphite_transform_loops): Provide opt-info for optimized nests.
26972 * tree-parloop.c (parallelize_loops): Provide opt-info for
26973 parallelized loops.
26975 2017-02-01 Richard Biener <rguenther@suse.de>
26977 PR middle-end/79315
26978 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
26979 was not set before.
26981 2017-02-01 Richard Biener <rguenther@suse.de>
26983 PR tree-optimization/71824
26984 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
26985 Verify the loops are valid in the merged SESE region.
26986 (scop_detection::can_represent_loop_1): Check analyzing the
26987 evolution of the number of iterations in the region succeeds.
26989 2017-01-31 Ian Lance Taylor <iant@golang.org>
26991 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
26992 REG_ARGS_SIZE note to 32-bit push insns and call insn.
26994 2017-01-31 David Malcolm <dmalcolm@redhat.com>
26996 PR preprocessor/79210
26997 * input.c (get_substring_ranges_for_loc): Replace line_width
26998 assertion with error-handling.
27000 2017-01-31 Richard Biener <rguenther@suse.de>
27002 PR tree-optimization/77318
27003 * graphite-sese-to-poly.c (extract_affine): Fix assert.
27004 (create_pw_aff_from_tree): Take loop parameter.
27005 (add_condition_to_pbb): Pass loop of the condition to
27006 create_pw_aff_from_tree.
27008 2017-01-31 Jakub Jelinek <jakub@redhat.com>
27010 * config/s390/s390.c (s390_asan_shadow_offset): New function.
27011 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
27013 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
27017 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
27019 (convert_int_to_float128): Likewise.
27020 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
27021 (convert_int_to_float128): Likewise.
27022 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
27023 (UNSPEC_IEEE128_CONVERT): Likewise.
27024 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
27025 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
27026 Use local variables for IBM extended format.
27027 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
27028 (fix_trunc<mode>si2_fprs): Likewise.
27029 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
27030 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
27031 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
27032 to know that we can now have integers of all sizes in vector
27034 (fix<uns>_<mode>di2_hw): Likewise.
27035 (float<uns>_<mode>si2_hw): Likewise.
27036 (fix_<mode>si2_hw): Likewise.
27037 (fixuns_<mode>si2_hw): Likewise.
27038 (float<uns>_<mode>di2_hw): Likewise.
27039 (float_<mode>di2_hw): Likewise.
27040 (float_<mode>si2_hw): Likewise.
27041 (floatuns_<mode>di2_hw): Likewise.
27042 (floatuns_<mode>si2_hw): Likewise.
27043 (xscvqp<su>wz_<mode>): Delete, no longer used.
27044 (xscvqp<su>dz_<mode>): Likewise.
27045 (xscv<su>dqp_<mode>): Likewise.
27046 (ieee128_mfvsrd_64bit): Likewise.
27047 (ieee128_mfvsrd_32bit): Likewise.
27048 (ieee128_mfvsrwz): Likewise.
27049 (ieee128_mtvsrw): Likewise.
27050 (ieee128_mtvsrd_64bit): Likewise.
27051 (ieee128_mtvsrd_32bit): Likewise.
27053 2017-01-31 Martin Liska <mliska@suse.cz>
27056 * ipa-prop.c (ipa_free_all_node_params): Call release method
27057 instead of ~sumbol_summary to not to trigger double times
27060 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
27062 PR tree-optimization/71691
27063 * bitmap.h (class auto_bitmap): New.
27064 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
27065 is_maybe_undefined instead of ssa_undefined_value_p.
27067 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27069 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
27070 __S390_ARCH_LEVEL__ to __ARCH__.
27072 2017-01-31 Jakub Jelinek <jakub@redhat.com>
27074 PR tree-optimization/79267
27075 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
27076 if should_remove_lhs_p is true.
27078 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
27081 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
27082 (add_alignment_attribute): New.
27083 (base_type_die): Add alignment attribute.
27084 (subrange_type_die): Likewise.
27085 (modified_type_die): Likewise.
27086 (gen_array_type_die): Likewise.
27087 (gen_descr_array_type_die: Likewise.
27088 (gen_enumeration_type_die): Likewise.
27089 (gen_subprogram_die): Likewise.
27090 (gen_variable_die): Likewise.
27091 (gen_field_die): Likewise.
27092 (gen_ptr_to_mbr_type_die): Likewise.
27093 (gen_struct_or_union_type_die): Likewise.
27094 (gen_subroutine_type_die): Likewise.
27095 (gen_typedef_die): Likewise.
27096 (base_type_cmp): Compare alignment attribute.
27098 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27101 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
27102 (setb_unsigned) New pattern for setb with CCUNS.
27103 * config/rs6000/rs6000.c (expand_block_compare): Use a different
27104 subfc./subfe sequence to avoid overflow problems. Generate a
27105 shorter sequence with cmpld/setb for power9.
27106 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
27107 for generating subfc. instruction.
27108 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
27109 now uses this instruction.
27111 2017-01-30 Ian Lance Taylor <iant@google.com>
27114 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
27115 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
27117 2017-01-30 Martin Sebor <msebor@redhat.com>
27119 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
27120 Move constant to the right of a relational operator.
27121 (get_mpfr_format_length, format_character, format_string): Ditto.
27122 (should_warn_p, maybe_warn): Same.
27124 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
27126 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
27129 * asan.c (get_translation_unit_decl): Remove function.
27130 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
27132 2017-01-30 Martin Liska <mliska@suse.cz>
27134 PR gcov-profile/79259
27135 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
27136 -fprofile-generate.
27138 2017-01-30 Martin Liska <mliska@suse.cz>
27141 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
27142 Initialize variables with NULL value.
27144 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
27147 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
27149 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
27151 2017-01-30 Richard Biener <rguenther@suse.de>
27153 PR tree-optimization/79276
27154 * tree-vrp.c (process_assert_insertions): Properly adjust common
27155 when removing a duplicate.
27157 * gcc.dg/torture/pr79276.c: New testcase.
27159 2017-01-30 Richard Biener <rguenther@suse.de>
27161 PR tree-optimization/79256
27162 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
27163 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
27165 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
27167 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
27170 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
27171 ("*r<noxa>sbg_<mode>_sll_bitmask")
27172 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
27173 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
27174 Use contiguous_bitmask_nowrap_operand.
27176 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27179 * config/rs6000/altivec.h (vec_xl): Revise #define.
27180 (vec_xst): Likewise.
27182 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
27184 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
27186 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
27188 PR rtl-optimization/79194
27189 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
27190 traps before call to bypass_conditional_jumps.
27192 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
27194 PR tree-optimization/71374
27195 * lra-constraints.c (check_conflict_input_operands): New.
27196 (match_reload): Use it.
27198 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
27201 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
27202 account to calculate conflict_set.
27204 2017-01-27 Bin Cheng <bin.cheng@arm.com>
27206 PR rtl-optimization/78559
27207 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
27208 other_insn in combine.
27210 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
27212 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
27213 uint16_type_node for BT_UINT16.
27215 2017-01-27 David Malcolm <dmalcolm@redhat.com>
27217 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
27218 "RTL Tests" to menu.
27219 (GIMPLE Tests): New node.
27220 (RTL Tests): New node.
27222 2017-01-27 Richard Biener <rguenther@suse.de>
27224 PR tree-optimization/79245
27225 * tree-loop-distribution.c (distribute_loop): Apply cost
27226 modeling also to detected patterns.
27228 2017-01-27 Richard Biener <rguenther@suse.de>
27230 PR tree-optimization/71433
27231 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
27232 (compare_assert_loc): New function.
27233 (process_assert_insertions): Sort and optimize assert locations
27234 to remove duplicates and push down identical assertions on
27235 edges to their destination block.
27237 2017-01-27 Richard Biener <rguenther@suse.de>
27239 PR tree-optimization/79244
27240 * tree-vrp.c (remove_range_assertions): Forcefully propagate
27241 out SSA names even if abnormal.
27243 2017-01-27 Jakub Jelinek <jakub@redhat.com>
27245 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
27246 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
27247 instead of MPFR_RNDN.
27249 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
27252 * arm.c (arm_option_override): Don't call build_target_option_node
27253 until after doing all option overrides.
27254 (arm_valid_target_attribute_tree): Likewise.
27256 2017-01-27 Martin Liska <mliska@suse.cz>
27258 * doc/invoke.texi (-fprofile-arcs): Document profiling support
27259 for {cd}tors and C++ {cd}tors.
27261 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
27263 * config/s390/s390.md ("*setmem_long_and")
27264 ("*setmem_long_and_31z"): Use zero_extend instead of and.
27266 2017-01-26 Martin Sebor <msebor@redhat.com>
27268 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
27271 2017-01-26 Martin Sebor <msebor@redhat.com>
27273 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
27274 HAVE_DFmode before using XFmode or DFmode.
27275 (parse_directive): Avoid using the z length modifier to avoid
27276 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
27278 PR middle-end/78703
27279 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
27280 to accept adjustment as an array.
27281 (get_int_range): New function.
27282 (struct directive): Make width and prec arrays.
27283 (directive::set_width, directive::set_precision): Call get_int_range.
27284 (format_integer, format_floating): Handle width and precision ranges.
27285 (format_string, parse_directive): Same.
27287 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27290 * dwarf2out.c (generate_skeleton_bottom_up): For children with
27291 comdat_type_p set, just clone them, but keep the children in the
27295 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
27296 which have direct callers with -fvar-tracking-assignments enabled
27298 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
27299 inside of type units.
27301 2017-01-26 Martin Sebor <msebor@redhat.com>
27303 PR middle-end/78703
27304 * gimple-ssa-sprintf.c (struct result_range): Add likely and
27306 (struct format_result): Replace number_chars, number_chars_min,
27307 and number_chars_max with a single member of struct result_range.
27309 (format_result::operator+=): Adjust.
27310 (struct fmtresult): Remove bounded. Handle likely and unlikely
27312 (fmtresult::adjust_for_width_or_precision): New function.
27313 (fmtresult:type_max_digits): New function.
27314 (bytes_remaining): Handle likely and unlikely counters.
27315 (min_bytes_remaining): Remove.
27316 (format_percent): Simplify.
27317 (format_integer, format_floating): Set likely and unlikely counters.
27318 (get_string_length, format_character, format_string): Same.
27319 (format_plain, should_warn_p): New function.
27320 (maybe_warn): Call should_warn_p. Update diagnostic messages
27321 and handle those for all directives, including plain strings.
27322 (format_directive): Handle likely and unlikely counters.
27323 Remove unnecessary quoting from diagnostics. Add an informational
27325 (add_bytes): Remove.
27326 (pass_sprintf_length::compute_format_length): Simplify.
27327 (try_substitute_return_value): Handle likely and unlikely counters.
27329 2017-01-26 Carl Love <cel@us.ibm.com>
27331 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
27332 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
27334 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
27337 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
27338 endianess for subregs into account.
27339 * lra-constraints.c (lra_constraints): Do risky transformations
27340 always on the first iteration.
27341 * lra-lives.c (check_pseudos_live_through_calls): Add arg
27342 last_call_used_reg_set.
27343 (process_bb_lives): Define and use last_call_used_reg_set.
27344 * lra.c (lra): Always continue after lra_constraints on the first
27347 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
27349 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
27351 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
27353 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27355 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
27356 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
27357 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
27358 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
27359 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
27360 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
27361 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
27362 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
27363 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
27365 2017-01-26 Marek Polacek <polacek@redhat.com>
27368 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
27369 for the third operand.
27371 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27373 PR middle-end/79236
27374 * omp-low.c (struct omp_context): Add simt_stmt field.
27375 (scan_omp_for): Return omp_context *.
27376 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
27377 context to the _simt_ SIMD stmt.
27378 (lower_omp_for): For combined SIMD with sibling _simt_
27379 SIMD, make sure to use the same decls in _looptemp_
27380 clauses as in the sibling.
27382 2017-01-26 David Sherwood <david.sherwood@arm.com>
27384 PR middle-end/79212
27385 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
27388 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27391 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
27392 emit fld b; fld a; if possible.
27394 * brig-builtins.def: Update copyright years.
27395 * config/arm/arm_acle_builtins.def: Update copyright years.
27397 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
27400 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
27401 constraint instead of o for the stxsd instruction.
27403 2017-01-25 Carl Love <cel@us.ibm.com>
27405 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
27406 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
27408 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
27410 * doc/invoke.texi (C++ Dialect Options): Fix typo.
27412 2017-01-25 Richard Biener <rguenther@suse.de>
27414 PR tree-optimization/69264
27415 * target.def (vector_alignment_reachable): Improve documentation.
27416 * doc/tm.texi: Regenerate.
27417 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
27419 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
27420 earlier changes with respect to TYPE_USER_ALIGN.
27421 (vector_alignment_reachable_p): Likewise. Improve dumping.
27423 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27426 * config/arm/arm.md (xordi3): Force constant operand into a register
27429 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27431 * doc/invoke.texi (-fstore-merging): Correct default optimization
27432 levels at which it is enabled.
27433 (-O): Move -fstore-merging from list to...
27436 2017-01-25 Richard Biener <rguenther@suse.de>
27439 * omp-expand.c: Include debug.h.
27440 (expand_omp_taskreg): Make sure to generate early debug before
27441 outlining anything from a function.
27442 (expand_omp_target): Likewise.
27443 (grid_expand_target_grid_body): Likewise.
27445 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
27448 * asan.c (get_translation_unit_decl): New function.
27449 (asan_add_global): Extract modules file name from globals
27450 TRANSLATION_UNIT_DECL name.
27452 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
27455 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
27456 for long calls with APCS frame and VFP.
27458 2017-01-24 David Malcolm <dmalcolm@redhat.com>
27460 * cfg.c (original_copy_tables_initialized_p): New function.
27461 * cfg.h (original_copy_tables_initialized_p): New decl.
27462 * cfgrtl.c (relink_block_chain): Guard the call to
27463 free_original_copy_tables with a call to
27464 original_copy_tables_initialized_p.
27465 * cgraph.h (symtab_node::native_rtl_p): New decl.
27466 * cgraphunit.c (symtab_node::native_rtl_p): New function.
27467 (symtab_node::needed_p): Don't assert for early assembly output
27468 for __RTL functions.
27469 (cgraph_node::finalize_function): Set "force_output" for __RTL
27471 (cgraph_node::analyze): Bail out early for __RTL functions.
27472 (analyze_functions): Update assertion to support __RTL functions.
27473 (cgraph_node::expand): Bail out early for __RTL functions.
27474 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
27476 * function.h (struct function): Update comment for field
27478 * gimple-expr.c: Include "tree-pass.h".
27479 (gimple_has_body_p): Return false for __RTL functions.
27480 * Makefile.in (OBJS): Add run-rtl-passes.o.
27481 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
27483 (gcc::pass_manager::get_clean_slate): New accessor.
27484 * passes.c: Include "insn-addr.h".
27485 (should_skip_pass_p): Add logging. Update logic for running
27486 "expand" to be compatible with both __GIMPLE and __RTL. Guard
27487 property-provider override so it is only done for gimple passes.
27489 (skip_pass): New function.
27490 (execute_one_pass): Call skip_pass when skipping passes.
27491 * read-md.c (md_reader::read_char): Support filtering
27492 the input to a subset of line numbers.
27493 (md_reader::md_reader): Initialize fields
27494 m_first_line and m_last_line.
27495 (md_reader::read_file_fragment): New function.
27496 * read-md.h (md_reader::read_file_fragment): New decl.
27497 (md_reader::m_first_line): New field.
27498 (md_reader::m_last_line): New field.
27499 * read-rtl-function.c (function_reader::create_function): Only
27500 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
27501 curr_properties. Set DECL_INITIAL to a dummy block.
27502 (read_rtl_function_body_from_file_range): New function.
27503 * read-rtl-function.h (read_rtl_function_body_from_file_range):
27505 * run-rtl-passes.c: New file.
27506 * run-rtl-passes.h: New file.
27508 2017-01-24 Jeff Law <law@redhat.com>
27510 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
27513 2017-01-24 Bin Cheng <bin.cheng@arm.com>
27515 PR tree-optimization/79159
27516 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
27517 (record_nonwrapping_iv): Improve boundary using above function if no
27518 value range information.
27520 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
27521 Martin Jambor <mjambor@suse.cz>
27523 * brig-builtins.def: New file.
27524 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
27525 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
27526 (DEF_HSAIL_SAT_BUILTIN): Likewise.
27527 (DEF_HSAIL_INTR_BUILTIN): Likewise.
27528 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
27529 * builtin-types.def (BT_INT8): New.
27530 (BT_INT16): Likewise.
27531 (BT_UINT8): Likewise.
27532 (BT_UINT16): Likewise.
27533 (BT_FN_ULONG): Likewise.
27534 (BT_FN_UINT_INT): Likewise.
27535 (BT_FN_UINT_ULONG): Likewise.
27536 (BT_FN_UINT_LONG): Likewise.
27537 (BT_FN_UINT_PTR): Likewise.
27538 (BT_FN_ULONG_PTR): Likewise.
27539 (BT_FN_INT8_FLOAT): Likewise.
27540 (BT_FN_INT16_FLOAT): Likewise.
27541 (BT_FN_UINT32_FLOAT): Likewise.
27542 (BT_FN_UINT16_FLOAT): Likewise.
27543 (BT_FN_UINT8_FLOAT): Likewise.
27544 (BT_FN_UINT64_FLOAT): Likewise.
27545 (BT_FN_UINT16_UINT32): Likewise.
27546 (BT_FN_UINT32_UINT16): Likewise.
27547 (BT_FN_UINT16_UINT16_UINT16): Likewise.
27548 (BT_FN_INT_PTR_INT): Likewise.
27549 (BT_FN_UINT_PTR_UINT): Likewise.
27550 (BT_FN_LONG_PTR_LONG): Likewise.
27551 (BT_FN_ULONG_PTR_ULONG): Likewise.
27552 (BT_FN_VOID_UINT64_UINT64): Likewise.
27553 (BT_FN_UINT8_UINT8_UINT8): Likewise.
27554 (BT_FN_INT8_INT8_INT8): Likewise.
27555 (BT_FN_INT16_INT16_INT16): Likewise.
27556 (BT_FN_INT_INT_INT): Likewise.
27557 (BT_FN_UINT_FLOAT_UINT): Likewise.
27558 (BT_FN_FLOAT_UINT_UINT): Likewise.
27559 (BT_FN_ULONG_UINT_UINT): Likewise.
27560 (BT_FN_ULONG_UINT_PTR): Likewise.
27561 (BT_FN_ULONG_ULONG_ULONG): Likewise.
27562 (BT_FN_UINT_UINT_UINT): Likewise.
27563 (BT_FN_VOID_UINT_PTR): Likewise.
27564 (BT_FN_UINT_UINT_PTR: Likewise.
27565 (BT_FN_UINT32_UINT64_PTR): Likewise.
27566 (BT_FN_INT_INT_UINT_UINT): Likewise.
27567 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
27568 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
27569 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
27570 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
27571 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
27572 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
27573 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
27574 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
27575 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
27576 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
27577 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
27578 * doc/frontends.texi: List BRIG FE.
27579 * doc/install.texi (Testing): Add BRIG tesring requirements.
27580 * doc/invoke.texi (Overall Options): Mention BRIG.
27581 * doc/standards.texi (Standards): Doucment BRIG HSA version.
27583 2017-01-24 Richard Biener <rguenther@suse.de>
27585 PR translation/79208
27586 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
27588 2017-01-24 Martin Jambor <mjambor@suse.cz>
27591 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
27592 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
27593 and known_contexts.
27595 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
27597 PR middle-end/79123
27598 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
27599 casts from signed to unsigned really don't have a range.
27601 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
27603 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
27604 GMP_RNDx for compatiblity.
27606 2017-01-24 Martin Liska <mliska@suse.cz>
27609 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
27610 that would prevent us to call alloca with -1 as argument.
27612 2017-01-24 Jakub Jelinek <jakub@redhat.com>
27614 * dwarf2out.c (output_compilation_unit_header, output_file_names):
27615 Avoid -Wformat-security warning.
27617 2017-01-23 Andrew Pinski <apinski@cavium.com>
27619 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
27622 2017-01-23 Martin Sebor <msebor@redhat.com>
27624 PR middle-end/78703
27625 * gimple-ssa-sprintf.c (warn_level): New global.
27626 (format_integer): Use it here and throughout the rest of the file.
27627 Use the same switch to compute sign as base.
27628 (maybe_warn): New function.
27629 (format_directive): Factor out warnings into maybe_warn.
27630 Add debugging output. Use warn_level.
27631 (add_bytes): Use warn_level.
27632 (pass_sprintf_length::compute_format_length): Add debugging output.
27633 (try_substitute_return_value): Same.
27634 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
27636 PR middle-end/78703
27637 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
27638 (struct fmtresult, format_integer, format_floating): Adjust.
27639 (fmtresult::fmtresult): Set max correctly in two argument ctor.
27640 (get_string_length, format_string,format_directive): Same.
27641 (pass_sprintf_length::compute_format_length): Same.
27642 (try_substitute_return_value): Simplify slightly.
27644 PR middle-end/78703
27645 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
27646 (fmtresult::operator+=): Outlined.
27647 (struct fmtresult): Add ctors.
27648 (struct conversion_spec): Rename...
27649 (struct directive): ...to this. Add and remove data members.
27650 (directive::set_width, directive::set_precision): New functions.
27651 (format_percent): Use fmtresult ctor.
27652 (get_width_and_precision): Remove.
27653 (format_integer): Make naming changes. Avoid computing width and
27655 (format_floating): Same. Adjust indentation.
27656 (format_character, format_none): New functions.
27657 (format_string): Moved character handling to format_character.
27658 (format_directive): Remove arguments, change return type.
27659 (parse_directive): New function.
27660 (pass_sprintf_length::compute_format_length): Move directive
27661 parsing to parse_directive.
27663 2017-01-23 Jakub Jelinek <jakub@redhat.com>
27665 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
27666 (assign_assembler_name_if_needed): ... this.
27667 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
27668 (assign_assembler_name_if_needed): ... this.
27669 (free_lang_data_in_cgraph): Adjust callers.
27670 * cgraphunit.c (cgraph_node::analyze): Likewise.
27671 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
27674 2017-01-23 Richard Biener <rguenther@suse.de>
27676 PR tree-optimization/79088
27677 PR tree-optimization/79188
27678 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
27679 resetting loop bounds after last path deletion. Reset loop
27680 bounds of the target loop, make code match the comments.
27681 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
27682 Make sure loops need no fixups.
27684 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
27686 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
27687 exponent support with double type for first argument.
27688 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
27689 type returned by __builtin_vec_extract_sig,
27690 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
27691 functions from "vector int" to "vector unsigned int" or from
27692 "vector long long int" to "vector unsigned long long int".
27693 Changed type returned by __builtin_vec_extract_exp,
27694 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
27695 functions from "vector int" to "vector unsigned int" or from
27696 "vector long long int" to "vector unsigned long long int".
27697 Changed return type of __builtin_vec_test_data_class,
27698 __builtin_vec_test_data_class_sp, and
27699 __builtin_vec_test_data_class_dp from "vector int" to
27700 "vector bool int" or from "vector long long int" to "vector bool
27701 long long int" and changed second argument type from "unsigned
27702 int" to "int". Added new overloaded function forms "vector float
27703 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
27704 "vector float __builtin_vec_insert_exp_sp (vector float, vector
27705 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
27706 double, vector unsigned long long int)" and "vector double
27707 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
27708 long int)". Changed return type of
27709 __builtin_scalar_test_data_class and
27710 __builtin_scalar_test_data_class_sp and
27711 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
27712 int" and changed second argument from "unsigned int" to "int".
27713 Changed type returned by __builtin_scalar_test_neg,
27714 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
27715 from "int" to "bool int". Added new overloaded function form
27716 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
27717 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
27718 exponent double-precision with floating point first argument.
27719 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
27720 documentation of scalar_test_data_class, scalar_test_neg,
27721 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
27722 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
27723 vec_test_data_class built-in functions to reflect refinements in
27724 their type signatures.
27726 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
27728 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
27730 (aarch64_elf_asm_destructor): Likewise.
27732 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
27734 PR rtl-optimization/78634
27735 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
27736 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
27737 * ifcvt.c (noce_try_cmove): Add missing cost check.
27739 PR rtl-optimization/71724
27740 * combine.c (if_then_else_cond): Look for situations where it is
27741 beneficial to undo the work of one of the recursive calls.
27743 2017-01-23 Bin Cheng <bin.cheng@arm.com>
27745 PR tree-optimization/70754
27746 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
27747 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
27748 combined stmt before it if not NULL.
27749 (combine_chains): Process refs reversely and compute dominance point
27752 2017-01-23 Martin Liska <mliska@suse.cz>
27754 PR tree-optimization/79196
27755 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
27756 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
27758 (strlen_optimize_stmt): Call the renamed function.
27760 2017-01-23 Michael Matz <matz@suse.de>
27762 PR tree-optimization/78384
27763 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
27765 2017-01-23 Richard Biener <rguenther@suse.de>
27767 PR tree-optimization/79186
27768 * tree-vrp.c (register_new_assert_for): Make sure we've seen
27769 both incoming edges before moving an assert.
27771 2017-01-23 Martin Jambor <mjambor@suse.cz>
27773 * ipa-prop.c (load_from_param_1): Removed.
27774 (load_from_unmodified_param): Bits from load_from_param_1 put back
27776 (load_from_param): Removed.
27777 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
27778 with stmt. Reverted back to use of load_from_unmodified_param.
27780 2017-01-23 Martin Jambor <mjambor@suse.cz>
27783 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
27784 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
27785 field a pointer to garbage collected vector, mark lattices and
27786 ipcp_orig_node with GTY((skip)).
27787 (ipa_get_param_count): Adjust to descriptors being a pointer.
27788 (ipa_get_param): Likewise.
27789 (ipa_get_type): Likewise.
27790 (ipa_get_param_move_cost): Likewise.
27791 (ipa_set_param_used): Likewise.
27792 (ipa_get_controlled_uses): Likewise.
27793 (ipa_set_controlled_uses): Likewise.
27794 (ipa_is_param_used): Likewise.
27795 (ipa_node_params_t): Move into garbage collector. New methods insert
27797 (ipa_node_params_sum): Annotate wth GTY(()).
27798 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
27800 (ipa_load_from_parm_agg): Adjust declaration.
27801 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
27802 * ipa-profile.c (ipa_profile): Likewise.
27803 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
27804 (ipa_populate_param_decls): Make descriptors parameter garbage
27806 (ipa_dump_param): Adjust to descriptors being a pointer.
27807 (ipa_alloc_node_params): Likewise.
27808 (ipa_initialize_node_params): Likewise.
27809 (load_from_param_1): Make descriptors parameter garbage collected.
27810 (load_from_unmodified_param): Likewise.
27811 (load_from_param): Likewise.
27812 (ipa_load_from_parm_agg): Likewise.
27813 (ipa_node_params::~ipa_node_params): Removed.
27814 (ipa_free_all_node_params): Remove call to delete operator.
27815 (ipa_node_params_t::insert): New.
27816 (ipa_node_params_t::remove): Likewise.
27817 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
27818 copy known_csts and known_contexts vectors.
27819 (ipa_read_node_info): Adjust to descriptors being a pointer.
27820 (ipcp_modif_dom_walker): Make m_descriptors field garbage
27822 (ipcp_transform_function): Make descriptors variable garbage
27825 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
27827 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
27828 * config/i386/avx512dqintrin.h: Ditto.
27829 * config/i386/avx512fintrin.h: Ditto.
27830 * config/i386/i386.c: Handle new builtins.
27831 * config/i386/i386-builtin.def: Add new builtins.
27832 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
27833 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
27835 2017-01-23 Jakub Jelinek <jakub@redhat.com>
27836 Martin Liska <mliska@suse.cz>
27838 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
27839 * asan.c (asan_expand_poison_ifn): Support stores and use
27840 appropriate ASAN report function.
27841 * internal-fn.c (expand_ASAN_POISON_USE): New function.
27842 * internal-fn.def (ASAN_POISON_USE): Declare.
27843 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
27844 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
27845 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
27846 ASAN_POISON calls w/o LHS.
27847 * tree-ssa.c (execute_update_addresses_taken): Create clobber
27848 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
27849 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
27850 * gimplify.c (asan_poison_variables): Add attribute
27851 use_after_scope_memory to variables that really needs to live
27853 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
27854 having the attribute.
27856 2017-01-23 Martin Liska <mliska@suse.cz>
27858 * asan.c (create_asan_shadow_var): New function.
27859 (asan_expand_poison_ifn): Likewise.
27860 * asan.h (asan_expand_poison_ifn): New declaration.
27861 * internal-fn.c (expand_ASAN_POISON): Likewise.
27862 * internal-fn.def (ASAN_POISON): New builtin.
27863 * sanopt.c (pass_sanopt::execute): Expand
27864 asan_expand_poison_ifn.
27865 * tree-inline.c (copy_decl_for_dup_finish): Make function
27867 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
27868 * tree-ssa.c (is_asan_mark_p): New function.
27869 (execute_update_addresses_taken): Rewrite local variables
27870 (identified just by use-after-scope as addressable) into SSA.
27872 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
27874 * doc/install.texi (Specific): opensource.apple.com uses https
27875 now. Remove trailing slash.
27877 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
27879 * README.Portability: Remove note on an Irix compatibility issue.
27881 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
27883 * gcov.c (INCLUDE_ALGORITHM): Define.
27884 (INCLUDE_VECTOR): Define.
27885 No longer include <vector> and <algorithm> directly.
27887 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
27889 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
27891 * doc/invoke.texi (Code Gen Options): Ditto.
27893 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
27896 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
27898 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
27900 rtl-optimization/79125
27901 * cprop.c (local_cprop_pass): Handle cases where we make an
27902 unconditional trap.
27904 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
27908 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
27909 read from, for big endian.
27911 2017-01-20 Jiong Wang <jiong.wang@arm.com>
27913 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
27914 register pauth builtins for LP64 only.
27916 2017-01-20 Marek Polacek <polacek@redhat.com>
27919 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
27922 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
27924 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
27926 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
27927 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
27928 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
27930 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27933 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
27934 in big-endian mode when they are not a single duplicated value.
27936 2017-01-20 Richard Biener <rguenther@suse.de>
27938 * BASE-VER: Bump to 7.0.1.
27940 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
27942 * omp-low.c (omplow_simd_context): New struct. Use it...
27943 (lower_rec_simd_input_clauses): ...here and...
27944 (lower_rec_input_clauses): ...here to hold common data. Adjust all
27945 references to idx, lane, max_vf, is_simt.
27947 2017-01-20 Graham Markall <graham.markall@embecosm.com>
27949 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
27952 2017-01-20 Martin Jambor <mjambor@suse.cz>
27954 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
27955 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
27957 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
27958 (GTFILES): Rename hsa.c to hsa-common.c.
27959 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
27960 * hsa-dump.c: Likewise.
27961 * hsa-gen.c: Likewise.
27962 * hsa-regalloc.c: Likewise.
27963 * ipa-hsa.c: Likewise.
27964 * omp-expand.c: Likewise.
27965 * omp-low.c: Likewise.
27966 * toplev.c: Likewise.
27968 2017-01-20 Marek Polacek <polacek@redhat.com>
27971 * doc/invoke.texi: Document -Wduplicated-branches.
27972 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
27973 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
27974 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
27975 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
27976 return 0 only when not OEP_LEXICOGRAPHIC.
27977 (fold_build_cleanup_point_expr): Use the expression
27978 location when building CLEANUP_POINT_EXPR.
27979 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
27980 * tree.c (add_expr): Handle error_mark_node.
27982 2017-01-20 Martin Liska <mliska@suse.cz>
27985 * tree-profile.c (init_ic_make_global_vars): Do not call
27987 (gimple_init_gcov_profiler): Likewise.
27989 2017-01-20 Martin Liska <mliska@suse.cz>
27992 * cgraph.h (maybe_create_reference): Remove argument and
27994 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
27996 * ipa-cp.c (create_specialized_node): Likewise.
27997 * symtab.c (symtab_node::maybe_create_reference): Handle
27998 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
28000 2017-01-20 Martin Liska <mliska@suse.cz>
28002 * read-rtl-function.c (function_reader::create_function): Use
28003 build_decl instread of build_decl_stat.
28005 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
28007 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
28008 * config/i386/avx512dqintrin.h: Ditto.
28009 * config/i386/avx512fintrin.h: Ditto.
28010 * config/i386/i386-builtin-types.def: Add new types.
28011 * config/i386/i386.c: Handle new types.
28012 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
28013 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
28014 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
28015 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
28016 (__builtin_ia32_kshiftridi): New.
28017 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
28019 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
28023 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
28024 define to rs6000_init_stack_protect_guard.
28025 (rs6000_init_stack_protect_guard): New function.
28027 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
28028 Yunqiang Su <yunqiang.su@imgtec.com>
28030 * config.gcc (supported_defaults): Add madd4.
28031 (with_madd4): Add validation.
28032 (all_defaults): Add madd4.
28033 * config/mips/mips.opt (mmadd4): New option.
28034 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
28036 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
28038 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
28039 (ISA_HAS_FUSED_MADD4): Likewise.
28040 * doc/invoke.texi (-mmadd4): Document the new option.
28041 * doc/install.texi (--with-madd4): Document the new option.
28043 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28045 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
28046 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
28047 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
28048 (aarch64_init_pauth_hint_builtins): New.
28049 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
28050 (aarch64_expand_builtin): Expand new builtins.
28052 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28054 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
28055 * combine-stack-adj.c (no_unhandled_cfa): Handle
28056 REG_CFA_TOGGLE_RA_MANGLE.
28057 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
28058 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
28059 info for return address signing.
28060 (aarch64_expand_epilogue): Likewise.
28062 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28064 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
28065 * config/aarch64/aarch64-protos.h
28066 (aarch64_return_address_signing_enabled): New declaration.
28067 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
28069 (aarch64_expand_prologue): Sign return address before it's pushed onto
28071 (aarch64_expand_epilogue): Authenticate return address fetched from
28073 (aarch64_override_options): Sanity check for ILP32 and ISA level.
28074 (aarch64_attributes): New function attributes for "sign-return-address".
28075 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
28076 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
28077 ("*do_return"): Generate combined instructions according to key index.
28078 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
28079 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
28081 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
28082 * config/aarch64/aarch64.opt (msign-return-address=): New.
28083 * doc/extend.texi (AArch64 Function Attributes): Documents
28084 "sign-return-address=".
28085 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
28087 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
28089 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
28090 overall option summary.
28092 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28094 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
28095 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
28096 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
28097 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
28099 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
28101 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
28102 -mpower9-minmax by default for -mcpu=power9.
28103 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
28104 128-bit floating point.
28106 2017-01-20 Alan Modra <amodra@gmail.com>
28108 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
28109 optimizing for size.
28111 2017-01-20 Alan Modra <amodra@gmail.com>
28114 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
28115 for strcmp and strncmp from corresponding builtin decl.
28117 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
28119 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
28120 instead of i386/rtems-64.h.
28121 * config/i386/rtems-64.h: Remove.
28123 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
28127 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
28129 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
28131 2017-01-19 Tamar Christina <tamar.christina@arm.com>
28133 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
28134 Change int to HOST_WIDE_INT.
28135 * config/aarch64/aarch64-protos.h
28136 (aarch64_simd_gen_const_vector_dup): Likewise.
28137 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
28139 2017-01-19 David Malcolm <dmalcolm@redhat.com>
28141 * langhooks-def.h (lhd_type_for_size): New decl.
28142 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
28143 * langhooks.c (lhd_type_for_size): New function, taken from
28146 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
28148 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
28149 define_bypass for CR latency.
28150 (power9-cracked-alu): Update bypass latency and remove power9-branch.
28151 (power9-alu2): Add define_bypass for CR latency.
28153 (power9-mul): Update insn latency.
28154 (power9-mul-compare): Update insn latency, bypass latency and remove
28157 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28159 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
28161 * config/aarch64/aarch64.md
28162 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
28163 aarch64_nopcrelative_literal_loads.
28164 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
28166 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
28168 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
28169 TARGET_LOONGSON_3A.
28170 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
28172 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
28175 * config.gcc (supported_defaults): Add lxc1-sxc1.
28176 (with_lxc1_sxc1): Add validation.
28177 (all_defaults): Add lxc1-sxc1.
28178 * config/mips/mips.opt (mlxc1-sxc1): New option.
28179 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
28181 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
28182 __mips_no_lxc1_sxc1.
28183 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
28184 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
28185 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
28187 2017-01-19 Richard Biener <rguenther@suse.de>
28189 PR tree-optimization/72488
28190 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
28191 sure to restore SSA info.
28192 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
28194 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
28196 PR rtl-optimization/79121
28197 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
28198 of the inner type when shifting an extended value.
28200 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
28203 * symtab.c (symtab_node::equal_address_to): Fix comparing of
28204 interposable aliases.
28206 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
28209 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
28210 Use the evmergelohi instruction.
28211 (mov_si<mode>_e500_subreg4_2_le): Likewise.
28212 (mov_sitf_e500_subreg8_2_be): Likewise.
28213 (mov_sitf_e500_subreg12_2_le): Likewise.
28214 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
28215 (mov_si<mode>_e500_subreg4_2_be): Likewise.
28216 (mov_sitf_e500_subreg8_2_le): Likewise.
28217 (mov_sitf_e500_subreg12_2_be): Likewise.
28219 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28221 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
28222 attribute from vecsimple to vecperm.
28223 (altivec_vbpermq2): Likewise.
28225 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28228 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
28230 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28231 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
28232 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
28233 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
28234 case where N arg is SIZE_MAX.
28235 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
28236 (cmpstrsi): Add pattern.
28238 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
28240 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
28241 __builtin_vec_revb builtins.
28242 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
28243 built-in functions to support generation of the ISA 3.0 XXBR<x>
28244 vector byte reverse instructions.
28245 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
28246 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
28247 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
28248 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
28249 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
28250 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
28251 (P9V_BUILTIN_VEC_REVB): Likewise.
28252 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
28253 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
28254 (p9_xxbrq_v16qi): Likewise.
28255 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
28256 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
28257 (p9_xxbrh_v8hi): Likewise.
28258 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
28259 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
28260 vec_revb built-in functions.
28262 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
28264 PR rtl-optimization/78952
28265 * config/i386/i386.md (any_extract): New code iterator.
28266 (*insvqi_2): Use any_extract for source operand.
28267 (*insvqi_3): Use any_shiftrt for source operand.
28269 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
28271 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
28273 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
28275 2017-01-18 Matthias Klose <doko@ubuntu.com>
28277 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
28279 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28281 * config/rs6000/altivec.h (vec_bperm): Change #define.
28282 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
28283 (altivec_vbpermq2): New define_insn.
28284 (altivec_vbpermd): Likewise.
28285 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
28286 function interface.
28287 (VBPERMD): Likewise.
28288 (VBPERM): New polymorphic function interface.
28289 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
28290 Add entries for P9V_BUILTIN_VEC_VBPERM.
28291 * doc/extend.texi: Add interfaces for vec_bperm.
28293 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28295 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
28296 first letter of error messages.
28297 (s390_resolve_overloaded_builtin): Likewise.
28298 * config/s390/s390.c (s390_expand_builtin): Likewise.
28299 (s390_invalid_arg_for_unprototyped_fn): Likewise.
28300 (s390_valid_target_attribute_inner_p): Likewise.
28301 * config/s390/s390.md ("tabort"): Likewise.
28303 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
28305 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
28306 (ISA_AVOID_DIV_HILO): New macro.
28307 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
28308 (ISA_HAS_DDIV): Likewise.
28310 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28312 * doc/invoke.texi (fabi-version): Correct number of occurrences.
28314 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28316 * doc/invoke.texi (fabi-version): Spelling fix.
28318 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28321 * doc/invoke.texi (fabi-version): Mention mangling fix for
28324 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28327 * doc/invoke.texi (fabi-version): Document discriminator mangling.
28329 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
28332 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
28333 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
28335 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
28337 (stack_protect_test): Ditto.
28338 * config/rs6000/rs6000.opt (mstack-protector-guard=,
28339 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
28341 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
28342 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
28343 -mstack-protector-guard-offset=.
28344 (RS/6000 and PowerPC Options): Ditto.
28346 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
28348 * config/i386/i386.h (MASK_CLASS_P): New define.
28349 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
28350 there are no registers from different register sets also when
28351 mask registers are used. Update function comment.
28352 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
28353 to (*k/*r) and (*k/*km) alternatives.
28355 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
28357 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
28358 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
28359 (EH_RETURN_HANDLER_RTX): New define.
28360 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
28361 Force frame pointer in EH return functions.
28362 (aarch64_expand_epilogue): Add barrier for eh_return.
28363 (aarch64_final_eh_return_addr): Remove.
28364 (aarch64_eh_return_handler_rtx): New function.
28365 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
28367 (aarch64_eh_return_handler_rtx): New prototype.
28369 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28371 * config/rs6000/altivec.h (vec_rlmi): New #define.
28372 (vec_vrlnm): Likewise.
28373 (vec_rlnm): Likewise.
28374 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
28375 (UNSPEC_VRLNM): Likewise.
28376 (VIlong): New mode iterator.
28377 (altivec_vrl<VI_char>mi): New define_insn.
28378 (altivec_vrl<VI_char>nm): Likewise.
28379 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
28381 (VRLDNM): Likewise.
28382 (RLNM): New polymorphic function entry.
28383 (VRLWMI): New monomorphic function entry.
28384 (VRLDMI): Likewise.
28385 (RLMI): New polymorphic function entry.
28386 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
28387 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
28388 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
28391 2017-01-17 Jakub Jelinek <jakub@redhat.com>
28394 * dwarf2out.c (field_byte_offset): Restore the
28395 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
28396 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
28397 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
28400 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
28403 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
28405 2017-01-17 Jakub Jelinek <jakub@redhat.com>
28408 * dwarf2out.c (add_data_member_location_attribute): For constant
28409 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
28410 instead of DW_AT_data_member_location, DW_AT_bit_offset and
28411 DW_AT_byte_size attributes.
28413 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
28415 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
28416 after forcing to constant memory when the code model is medium.
28418 2017-01-17 Julia Koval <julia.koval@intel.com>
28421 * config/i386/avx512fintrin.h
28422 (_mm512_i32gather_ps): Change __addr type to void const*.
28423 (_mm512_mask_i32gather_ps): Ditto.
28424 (_mm512_i32gather_pd): Ditto.
28425 (_mm512_mask_i32gather_pd): Ditto.
28426 (_mm512_i64gather_ps): Ditto.
28427 (_mm512_mask_i64gather_ps): Ditto.
28428 (_mm512_i64gather_pd): Ditto.
28429 (_mm512_mask_i64gather_pd): Ditto.
28430 (_mm512_i32gather_epi32): Ditto.
28431 (_mm512_mask_i32gather_epi32): Ditto.
28432 (_mm512_i32gather_epi64): Ditto.
28433 (_mm512_mask_i32gather_epi64): Ditto.
28434 (_mm512_i64gather_epi32): Ditto.
28435 (_mm512_mask_i64gather_epi32): Ditto.
28436 (_mm512_i64gather_epi64): Ditto.
28437 (_mm512_mask_i64gather_epi64): Ditto.
28438 (_mm512_i32scatter_ps): Change __addr type to void*.
28439 (_mm512_mask_i32scatter_ps): Ditto.
28440 (_mm512_i32scatter_pd): Ditto.
28441 (_mm512_mask_i32scatter_pd): Ditto.
28442 (_mm512_i64scatter_ps): Ditto.
28443 (_mm512_mask_i64scatter_ps): Ditto.
28444 (_mm512_i64scatter_pd): Ditto.
28445 (_mm512_mask_i64scatter_pd): Ditto.
28446 (_mm512_i32scatter_epi32): Ditto.
28447 (_mm512_mask_i32scatter_epi32): Ditto.
28448 (_mm512_i32scatter_epi64): Ditto.
28449 (_mm512_mask_i32scatter_epi64): Ditto.
28450 (_mm512_i64scatter_epi32): Ditto.
28451 (_mm512_mask_i64scatter_epi32): Ditto.
28452 (_mm512_i64scatter_epi64): Ditto.
28453 (_mm512_mask_i64scatter_epi64): Ditto.
28454 * config/i386/avx512pfintrin.h
28455 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
28456 (_mm512_mask_prefetch_i32gather_ps): Ditto.
28457 (_mm512_mask_prefetch_i64gather_pd): Ditto.
28458 (_mm512_mask_prefetch_i64gather_ps): Ditto.
28459 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
28460 (_mm512_prefetch_i32scatter_ps): Ditto.
28461 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
28462 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
28463 (_mm512_prefetch_i64scatter_pd): Ditto.
28464 (_mm512_prefetch_i64scatter_ps): Ditto.
28465 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
28466 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
28467 * config/i386/avx512vlintrin.h
28468 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
28469 (_mm_mmask_i32gather_ps): Ditto.
28470 (_mm256_mmask_i32gather_pd): Ditto.
28471 (_mm_mmask_i32gather_pd): Ditto.
28472 (_mm256_mmask_i64gather_ps): Ditto.
28473 (_mm_mmask_i64gather_ps): Ditto.
28474 (_mm256_mmask_i64gather_pd): Ditto.
28475 (_mm_mmask_i64gather_pd): Ditto.
28476 (_mm256_mmask_i32gather_epi32): Ditto.
28477 (_mm_mmask_i32gather_epi32): Ditto.
28478 (_mm256_mmask_i32gather_epi64): Ditto.
28479 (_mm_mmask_i32gather_epi64): Ditto.
28480 (_mm256_mmask_i64gather_epi32): Ditto.
28481 (_mm_mmask_i64gather_epi32): Ditto.
28482 (_mm256_mmask_i64gather_epi64): Ditto.
28483 (_mm_mmask_i64gather_epi64): Ditto.
28484 (_mm256_i32scatter_ps): Change __addr type to void*.
28485 (_mm256_mask_i32scatter_ps): Ditto.
28486 (_mm_i32scatter_ps): Ditto.
28487 (_mm_mask_i32scatter_ps): Ditto.
28488 (_mm256_i32scatter_pd): Ditto.
28489 (_mm256_mask_i32scatter_pd): Ditto.
28490 (_mm_i32scatter_pd): Ditto.
28491 (_mm_mask_i32scatter_pd): Ditto.
28492 (_mm256_i64scatter_ps): Ditto.
28493 (_mm256_mask_i64scatter_ps): Ditto.
28494 (_mm_i64scatter_ps): Ditto.
28495 (_mm_mask_i64scatter_ps): Ditto.
28496 (_mm256_i64scatter_pd): Ditto.
28497 (_mm256_mask_i64scatter_pd): Ditto.
28498 (_mm_i64scatter_pd): Ditto.
28499 (_mm_mask_i64scatter_pd): Ditto.
28500 (_mm256_i32scatter_epi32): Ditto.
28501 (_mm256_mask_i32scatter_epi32): Ditto.
28502 (_mm_i32scatter_epi32): Ditto.
28503 (_mm_mask_i32scatter_epi32): Ditto.
28504 (_mm256_i32scatter_epi64): Ditto.
28505 (_mm256_mask_i32scatter_epi64): Ditto.
28506 (_mm_i32scatter_epi64): Ditto.
28507 (_mm_mask_i32scatter_epi64): Ditto.
28508 (_mm256_i64scatter_epi32): Ditto.
28509 (_mm256_mask_i64scatter_epi32): Ditto.
28510 (_mm_i64scatter_epi32): Ditto.
28511 (_mm_mask_i64scatter_epi32): Ditto.
28512 (_mm256_i64scatter_epi64): Ditto.
28513 (_mm256_mask_i64scatter_epi64): Ditto.
28514 (_mm_i64scatter_epi64): Ditto.
28515 (_mm_mask_i64scatter_epi64): Ditto.
28516 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
28517 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
28518 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
28519 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
28520 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
28521 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
28522 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
28523 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
28524 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
28525 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
28526 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
28527 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
28528 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
28529 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
28530 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
28531 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
28532 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
28533 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
28534 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
28535 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
28536 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
28537 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
28538 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
28539 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
28540 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
28541 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
28542 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
28543 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
28544 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
28545 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
28546 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
28547 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
28548 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
28549 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
28550 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
28551 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
28552 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
28553 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
28554 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
28555 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
28556 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
28557 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
28558 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
28559 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
28560 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
28561 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
28562 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
28563 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
28564 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
28565 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
28566 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
28567 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
28568 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
28569 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
28570 definitions accordingly.
28572 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
28573 Kuan-Lin Chen <kuanlinchentw@gmail.com>
28576 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
28579 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
28582 * ira-conflicts.c (ira_build_conflicts): Update total conflict
28583 hard regs for inner regno.
28585 2017-01-17 Martin Liska <mliska@suse.cz>
28588 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
28589 assumption and add comment.
28591 2017-01-17 Nathan Sidwell <nathan@acm.org>
28593 * ipa-visibility.c (localize_node): New function, broken out of ...
28594 (function_and_variable_visibility): ... here. Call it.
28596 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
28598 PR middle-end/77445
28599 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
28600 correctly set frequency of oudgoing edge.
28601 (duplicate_thread_path): Fix profile updating.
28603 2017-01-17 Jakub Jelinek <jakub@redhat.com>
28606 * configure.ac: Add GCC_BASE_VER.
28607 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
28608 version from BASE-VER file.
28609 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
28610 (gcc.o): Depend on $(BASEVER).
28611 * common.opt (dumpfullversion): New option.
28612 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
28613 * doc/invoke.texi: Document -dumpfullversion.
28614 * doc/install.texi: Document --with-gcc-major-version-only.
28615 * configure: Regenerated.
28617 2017-01-17 Richard Biener <rguenther@suse.de>
28619 PR tree-optimization/71433
28620 * tree-vrp.c (register_new_assert_for): Merge same asserts
28621 on all incoming edges.
28622 (process_assert_insertions_for): Handle insertions at the
28625 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
28627 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
28628 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
28630 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
28633 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
28636 2017-01-17 Alan Modra <amodra@gmail.com>
28639 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
28640 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
28641 symbolic stack limit when pic.
28643 2017-01-16 Martin Sebor <msebor@redhat.com>
28645 PR tree-optimization/78608
28646 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
28648 2017-01-16 Jeff Law <law@redhat.com>
28651 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
28652 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
28653 for several include directories that may be relative to sysroot.
28654 * config/i386/x-mingw32 (gplus_includedir): Define.
28655 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
28656 (native_system_includedir): Likewise.
28657 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
28658 override if TARGET_SYSTEM_ROOT is defined.
28659 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
28661 PR tree-optimization/79090
28662 PR tree-optimization/33562
28663 PR tree-optimization/61912
28664 PR tree-optimization/77485
28665 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
28666 and computed trims into the dump file.
28668 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
28670 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
28672 2017-01-16 Jakub Jelinek <jakub@redhat.com>
28675 * gimplify.c (gimplify_init_constructor): If want_value and
28676 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
28680 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
28681 sequence. Formatting fixes.
28682 (doloop_optimize): Formatting fixes.
28685 * gcc.c (debug_level_greater_than_spec_func): New function.
28686 (static_spec_functions): Add debug-level-gt spec function.
28687 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
28689 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
28690 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
28691 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
28692 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
28693 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
28694 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
28696 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
28698 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
28699 QImode fixups to general and mask registers only.
28701 2017-01-16 Carl Love <cel@us.ibm.com>
28703 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
28704 for built-in functions
28705 vector signed char vec_nabs (vector signed char)
28706 vector signed short vec_nabs (vector signed short)
28707 vector signed int vec_nabs (vector signed int)
28708 vector signed long long vec_nabs (vector signed long long)
28709 vector float vec_nabs (vector float)
28710 vector double vec_nabs (vector double)
28711 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
28713 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
28714 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
28715 * doc/extend.texi: Update the documentation file for the new built-in
28718 2017-01-16 Martin Sebor <msebor@redhat.com>
28720 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
28723 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28725 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
28726 UNSPEC_VSX__XXSPLTD to require special splat handling.
28728 2017-01-16 David Malcolm <dmalcolm@redhat.com>
28731 * system.h: Poison strndup.
28733 2017-01-16 Alan Modra <amodra@gmail.com>
28736 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
28739 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
28741 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
28743 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
28745 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
28746 call recog here. Assert that INSN_CODE (insn) is non-negative.
28748 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
28751 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
28753 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
28754 in the currently scheduled RTL fragment.
28756 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
28758 PR rtl-optimization/78751
28759 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
28762 2017-01-14 Jeff Law <law@redhat.com>
28764 PR tree-optimization/79090
28765 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
28766 variable length stores.
28767 (compute_trims): Delete dead assignment to *trim_tail.
28768 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
28771 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
28773 PR rtl-optimization/78626
28774 PR rtl-optimization/78727
28775 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
28776 of a block, and split such blocks after everything else is finished.
28778 2017-01-14 Alan Modra <amodra@gmail.com>
28781 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
28782 target legitimate_combined_insn.
28783 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
28784 (rs6000_legitimate_combined_insn): New function.
28785 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
28787 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
28788 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
28789 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
28791 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
28793 * doc/frontends.texi (G++ and GCC): Remove references to Java.
28795 2017-01-13 Jeff Law <law@redhat.com>
28797 PR tree-optimization/33562
28798 PR tree-optimization/61912
28799 PR tree-optimization/77485
28800 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
28802 (delete_dead_assignment): Likewise.
28803 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
28804 statement to delete_dead_call and delete_dead_assignment.
28806 2017-01-13 David Malcolm <dmalcolm@redhat.com>
28809 * substring-locations.c (format_warning_va): Strengthen case 1 so
28810 that both endpoints of the substring must be within the format
28811 range for just the substring to be printed.
28813 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
28815 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
28816 * config/i386/i386.c (ix86_target_string): Add missing options
28817 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
28818 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
28819 flags_other and ix86_target_other to flags2_other. Display unknown
28821 (ix86_valid_target_attribute_inner_p): Add missing options and
28822 reorder options by implied ISAs, as in ix86_target_string.
28824 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
28826 * hash-table.h (hash_table::too_empty_p): New function.
28827 (hash_table::expand): Use it.
28828 (hash_table::traverse): Likewise.
28829 (hash_table::empty_slot): Use sizeof (value_type) instead of
28830 sizeof (PTR) to convert bytes to elements. Shrink the table
28831 if the current size is excessive for the current number of
28834 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
28836 * ira-costs.c (record_reg_classes): Break from the inner loop
28837 early once alt_fail is known to be true. Update outer loop
28838 handling accordingly.
28840 2017-01-13 Jeff Law <law@redhat.com>
28842 * tree-ssa-dse.c (decrement_count): New function.
28843 (increment_start_addr, maybe_trim_memstar_call): Likewise.
28844 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
28845 when we know the partially dead statement is a mem* function.
28847 PR tree-optimization/61912
28848 PR tree-optimization/77485
28849 * tree-ssa-dse.c: Include expr.h.
28850 (maybe_trim_constructor_store): New function.
28851 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
28853 PR tree-optimization/33562
28854 PR tree-optimization/61912
28855 PR tree-optimization/77485
28856 * doc/invoke.texi: Document new dse-max-object-size param.
28857 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
28858 * tree-ssa-dse.c: Include params.h.
28859 (dse_store_status): New enum.
28860 (initialize_ao_ref_for_dse): New, partially extracted from
28862 (valid_ao_ref_for_dse, normalize_ref): New.
28863 (setup_live_bytes_from_ref, compute_trims): Likewise.
28864 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
28865 (maybe_trim_partially_dead_store): Likewise.
28866 (maybe_trim_complex_store): Likewise.
28867 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
28868 Track what bytes live from the original store. Return tri-state
28869 for dead, partially dead or live.
28870 (dse_dom_walker): Add constructor, destructor and new private members.
28871 (delete_dead_call, delete_dead_assignment): New extracted from
28873 (dse_optimize_stmt): Make a member of dse_dom_walker.
28874 Use initialize_ao_ref_for_dse.
28876 PR tree-optimization/33562
28877 PR tree-optimization/61912
28878 PR tree-optimization/77485
28879 * sbitmap.h (bitmap_count_bits): Prototype.
28880 (bitmap_clear_range, bitmap_set_range): Likewise.
28881 * sbitmap.c (bitmap_clear_range): New function.
28882 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
28884 2017-01-13 Martin Liska <mliska@suse.cz>
28887 * function.c (set_cfun): Add new argument force.
28888 * function.h (set_cfun): Likewise.
28889 * ipa-inline-transform.c (inline_call): Use the function when
28890 strict alising from is dropped for function we inline to.
28892 2017-01-13 Richard Biener <rguenther@suse.de>
28894 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
28895 for dumping GIMPLE INTEGER_CSTs.
28897 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28899 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
28900 to 201112L since C++17.
28902 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
28905 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
28906 if -fsanitize=kernel-address is present.
28908 2017-01-13 Richard Biener <rguenther@suse.de>
28910 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
28911 as _Literal ( type ) number in case usual suffixes do not
28912 preserve all information.
28914 2017-01-13 Richard Biener <rguenther@suse.de>
28916 PR tree-optimization/77283
28917 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
28918 and ssa-iterators.h.
28919 (is_feasible_trace): Implement a cost model based on joiner
28922 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
28925 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
28926 char or short to __float128/_Float128 directly.
28928 2017-01-12 Martin Sebor <msebor@redhat.com>
28930 to -Wformat-overflow.
28931 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
28932 (min_bytes_remaining): Same.
28933 (get_string_length): Same.
28934 (format_string): Same.
28935 (format_directive): Same.
28937 (pass_sprintf_length::handle_gimple_call): Same.
28939 2017-01-12 Jakub Jelinek <jakub@redhat.com>
28941 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
28942 info.nowrite calls with no lhs that can't throw. Return bool
28943 whether gsi_remove has been called or not.
28944 (pass_sprintf_length::handle_gimple_call): Return bool whether
28945 try_substitute_return_value called gsi_remove. Formatting fix.
28946 (pass_sprintf_length::execute): Don't use gsi_remove if
28947 handle_gimple_call returned true.
28950 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
28951 be removed due to side-effects, don't remove following barrier nor
28952 turn the successor edge into fallthru edge.
28954 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28957 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
28958 element-reversing loads and stores as not swappable.
28960 2017-01-12 Nathan Sidwell <nathan@acm.org>
28961 Nicolai Stange <nicstange@gmail.com>
28963 * combine.c (try_combine): Don't ignore result of overlap checking
28964 loop. Combine overlap & asm check into single loop.
28966 2017-01-12 Richard Biener <rguenther@suse.de>
28968 * tree-pretty-print.c (dump_generic_node): Provide -gimple
28969 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
28971 2017-01-12 Richard Biener <rguenther@suse.de>
28973 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
28974 and TS_TARGET_OPTION directly derive from TS_BASE.
28975 * tree-core.h (tree_optimization_option): Derive from tree_base.
28976 (tree_target_option): Likewise.
28978 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
28980 * config/i386/i386.c (memory_address_length): Increase len
28981 only when rip_relative_addr_p returns false.
28983 2017-01-11 Julia Koval <julia.koval@intel.com>
28985 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
28986 (OPTION_MASK_ISA_SGX_SET): New.
28987 (ix86_handle_option): Handle OPT_msgx.
28988 * config.gcc: Added sgxintrin.h.
28989 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
28990 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
28991 * config/i386/i386.c (ix86_target_string): Add -msgx.
28993 (ix86_option_override_internal): Handle new options.
28994 (ix86_valid_target_attribute_inner_p): Add sgx.
28995 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
28996 * config/i386/i386.opt: Add msgx.
28997 * config/i386/sgxintrin.h: New file.
28998 * config/i386/x86intrin.h: Add sgxintrin.h.
29000 2017-01-11 Jakub Jelinek <jakub@redhat.com>
29003 * fold-const.c (maybe_nonzero_address): Return 1 for function
29005 (tree_single_nonzero_warnv_p): Don't handle function local objects
29009 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
29012 2017-01-11 David Malcolm <dmalcolm@redhat.com>
29015 * opts.c: Include "spellcheck.h"
29016 (struct string_fragment): New struct.
29017 (struct edit_distance_traits<const string_fragment &>): New
29019 (get_closest_sanitizer_option): New function.
29020 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
29022 2017-01-11 Jakub Jelinek <jakub@redhat.com>
29024 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
29026 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
29027 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
29028 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
29029 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
29030 for initial die_offset if dwarf_split_debug_info.
29031 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
29032 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
29034 (output_skeleton_debug_sections): Formatting fix. Use
29035 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
29036 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
29038 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
29040 * config/arm/cortex-a53.md: Add bypasses for
29041 cortex_a53_r2f_cvt.
29042 (cortex_a53_r2f): Only use for transfers.
29043 (cortex_a53_f2r): Likewise.
29044 (cortex_a53_r2f_cvt): Add reservation for conversions.
29045 (cortex_a53_f2r_cvt): Likewise.
29047 2017-01-11 Tamar Christina <tamar.christina@arm.com>
29049 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
29050 to all inlined functions, change static to extern.
29052 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
29055 * config/arm/arm.c (legitimize_pic_address): Handle reference to
29057 (arm_assemble_integer): Likewise.
29059 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
29061 * config.gcc: Use new awk script to check CPU, FPU and architecture
29062 parameters for --with-... options.
29063 * config/arm/parsecpu.awk: New file
29064 * config/arm/arm-cpus.in: New file.
29065 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
29067 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
29069 * config/arm/t-arm: Update dependency rules.
29070 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
29071 of processing .def files.
29072 * config/arm/genopt.sh: Deleted.
29073 * config/arm/gentune.sh: Deleted.
29074 * config/arm/arm-cores.def: Deleted.
29075 * config/arm/arm-arches.def: Deleted.
29076 * config/arm/arm-fpus.def: Deleted.
29077 * config/arm/arm-tune.md: Regenerated.
29078 * config/arm/arm-tables.opt: Regenerated.
29079 * config/arm/arm-cpu.h: New generated file.
29080 * config/arm/arm-cpu-data.h: New generated file.
29081 * config/arm/arm-cpu-cdata.h: New generated file.
29083 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
29086 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
29088 (input_varpool_node): Unpack dynamically_initialized bit.
29090 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
29092 PR rtl-optimization/79032
29093 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
29094 the alignment of the adjusted memory reference against that of MODE,
29095 instead of the alignment of the original memory reference.
29097 2017-01-11 Martin Jambor <mjambor@suse.cz>
29099 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
29101 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
29102 decorated functions.
29104 2017-01-11 Richard Biener <rguenther@suse.de>
29106 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
29107 set range/nonnull info for PHI results. Do not set it on
29108 stmts marked for removal.
29110 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
29112 * expr.c (store_field): In the bitfield case, fetch the return value
29113 from the registers before applying a single big-endian adjustment.
29114 Always do a final load for a BLKmode value not larger than a word.
29116 2017-01-10 David Malcolm <dmalcolm@redhat.com>
29119 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
29120 that we correctly handle column numbers greater than
29121 LINE_MAP_MAX_COLUMN_NUMBER.
29123 2017-01-10 Martin Sebor <msebor@redhat.com>
29125 PR middle-end/78245
29126 * gimple-ssa-sprintf.c (get_destination_size): Call
29127 {init,fini}object_sizes.
29128 * tree-object-size.c (addr_object_size): Adjust.
29129 (pass_through_call): Adjust.
29130 (pass_object_sizes::execute): Adjust.
29131 * tree-object-size.h (fini_object_sizes): Declare.
29133 2017-01-10 Martin Sebor <msebor@redhat.com>
29135 PR tree-optimization/78775
29136 * builtins.c (get_size_range): Move...
29137 * calls.c: ...to here.
29138 (alloc_max_size): Accept zero argument.
29139 (operand_signed_p): Remove.
29140 (maybe_warn_alloc_args_overflow): Call get_size_range.
29141 * calls.h (get_size_range): Declare.
29143 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
29145 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
29146 from TI's devices.csv file as of September 2016.
29147 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
29149 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
29151 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
29152 * doc/invoke.texi: Likewise.
29153 * doc/md.texi: Likewise.
29154 * doc/objc.texi: Likewise.
29156 2017-01-10 Joshua Conner <joshconner@google.com>
29158 * config/arm/fuchsia-elf.h: New file.
29159 * config/fuchsia.h: New file.
29160 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
29161 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
29163 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
29165 2016-01-10 Richard Biener <rguenther@suse.de>
29167 PR tree-optimization/79034
29168 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
29169 Propagate out degenerate PHIs in the joiner.
29171 2017-01-10 Martin Liska <mliska@suse.cz>
29173 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
29174 (sort_congruence_classes_by_decl_uid): Likewise.
29175 (sort_congruence_class_groups_by_decl_uid): Likewise.
29176 (sem_item_optimizer::merge_classes): Sort class, groups in these
29177 classes and members in the groups by DECL_UID of declarations.
29178 This would make merge operations stable.
29180 2017-01-10 Martin Liska <mliska@suse.cz>
29182 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
29183 usage of m_classes_vec.
29184 (sem_item_optimizer::~sem_item_optimizer): Likewise.
29185 (sem_item_optimizer::get_group_by_hash): Likewise.
29186 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
29187 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
29188 (sem_item_optimizer::verify_classes): Likewise.
29189 (sem_item_optimizer::process_cong_reduction): Likewise.
29190 (sem_item_optimizer::dump_cong_classes): Likewise.
29191 (sem_item_optimizer::merge_classes): Likewise.
29192 * ipa-icf.h (congruence_class_hash): Rename from
29193 congruence_class_group_hash. Remove declaration of m_classes_vec.
29195 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
29197 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
29198 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
29199 * config.gcc: Add avx512vpopcntdqintrin.h.
29200 * config/i386/avx512vpopcntdqintrin.h: New.
29201 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
29202 * config/i386/i386-builtin-types.def: Add new types.
29203 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
29204 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
29205 __builtin_ia32_vpopcountq_v8di_mask): New.
29206 * config/i386/i386-c.c (ix86_target_macros_internal): Define
29207 __AVX512VPOPCNTDQ__.
29208 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
29209 (PTA_AVX512VPOPCNTDQ): Define.
29210 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
29211 TARGET_AVX512VPOPCNTDQ_P): Define.
29212 * config/i386/i386.opt: Add mavx512vpopcntdq.
29213 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
29214 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
29216 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29218 PR middle-end/77484
29219 * predict.def (PRED_CALL): Set to 67.
29221 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
29223 * expr.c (store_field): In the bitfield case, if the value comes from
29224 a function call and is of an aggregate type returned in registers, do
29225 not modify the field mode; extract the value in all cases if the mode
29226 is BLKmode and the size is not larger than a word.
29228 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
29231 * config/i386/cpuid.h: Fix undefined behavior.
29233 2017-01-04 Jeff Law <law@redhat.com>
29235 PR tree-optimization/79007
29236 PR tree-optimization/67955
29237 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
29238 conservative for pt.null when flag_non_call_exceptions is on.
29240 2017-01-09 Jakub Jelinek <jakub@redhat.com>
29242 PR translation/79019
29243 PR translation/79020
29244 * params.def (PARAM_INLINE_MIN_SPEEDUP,
29245 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
29246 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
29248 * config/avr/avr.opt (maccumulate-args): Likewise.
29249 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
29250 * common.opt (freport-bug): Likewise.
29251 * cif-code.def (CIF_FINAL_ERROR): Likewise.
29252 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
29253 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
29254 translatable string.
29255 * config/i386/i386.c (function_value_32): Likewise.
29256 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
29257 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
29259 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
29260 * common/config/msp430/msp430-common.c (msp430_handle_option):
29262 * symtab.c (symtab_node::verify_base): Likewise.
29263 * opts.c (set_debug_level): Likewise.
29264 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
29265 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
29266 missing whitespace to translatable strings.
29267 * config/avr/avr.md (bswapsi2): Fix typo in comment.
29268 * config/sh/superh.h: Likewise.
29269 * config/i386/xopintrin.h: Likewise.
29270 * config/i386/znver1.md: Likewise.
29271 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
29272 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
29273 * double-int.h (struct double_int): Likewise.
29274 * double-int.c (div_and_round_double): Likewise.
29275 * wide-int.cc: Likewise.
29276 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
29277 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
29278 * cfgcleanup.c (crossjumps_occured): Renamed to ...
29279 (crossjumps_occurred): ... this.
29280 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
29283 PR tree-optimization/78899
29284 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
29285 returning bool return struct loop *, NULL for failure and the new
29287 (versionable_outer_loop_p): Don't version outer loop if it has
29288 dont_vectorized bit set.
29289 (tree_if_conversion): When versioning outer loop, ensure
29290 tree_if_conversion is performed also on the inner loop of the
29291 non-vectorizable outer loop copy.
29292 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
29293 LOOP_VECTORIZED in inner loop of the scalar outer loop and
29294 prevent vectorization of it.
29295 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
29296 the outer loop vectorization of the non-scalar version is attempted
29297 before vectorization of the inner loop in scalar version. If
29298 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
29299 vectorization of its inner loop.
29300 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
29301 has 2 inner loops, rename also on edges from bb whose single pred
29302 is outer_loop->header. Fix typo in function comment.
29304 2017-01-09 Martin Sebor <msebor@redhat.com>
29307 * asan.c (asan_emit_stack_protection): Increase local buffer size
29308 to avoid snprintf truncation warning.
29310 2017-01-09 Andrew Pinski <apinski@cavium.com>
29312 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
29313 to reference thunderx2t99 for the tuning structure
29314 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
29316 (thunderx2t99_extra_costs): This.
29317 * config/aarch64/aarch64-tune.md: Regenerate.
29318 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
29319 (vulcan_addrcost_table): This.
29320 (vulcan_regmove_cost): Rename to ...
29321 (thunderx2t99_regmove_cost): This.
29322 (vulcan_vector_cost): Rename to ...
29323 (thunderx2t99_vector_cost): this.
29324 (vulcan_branch_cost): Rename to ...
29325 (thunderx2t99_branch_cost): This.
29326 (vulcan_tunings): Rename to ...
29327 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
29328 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
29330 2017-01-09 Martin Jambor <mjambor@suse.cz>
29334 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
29335 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
29336 (propagate_vr_accross_jump_function): Use the above function for all
29337 value range computations for pass-through jump functions and type
29338 converasion from explicit value range values.
29339 (ipcp_propagate_stage): Do not attempt to deduce types of formal
29340 parameters from TYPE_ARG_TYPES.
29341 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
29342 (ipa_write_node_info): Stream type of the actual argument.
29343 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
29345 2017-01-09 Martin Liska <mliska@suse.cz>
29348 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
29349 (lookup_compiler): Do not show error message with have_E.
29351 2017-01-09 Jakub Jelinek <jakub@redhat.com>
29353 PR tree-optimization/78938
29354 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
29355 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
29356 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
29357 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
29360 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29362 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
29365 2017-01-09 Richard Biener <rguenther@suse.de>
29367 PR tree-optimization/78997
29368 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
29369 name condition properly.
29371 2017-01-09 Richard Biener <rguenther@suse.de>
29374 * dwarf2out.c (is_cxx): New overload with context.
29375 (is_naming_typedef_decl): Use it.
29377 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
29379 * invoke.texi (Option Summary): Correct spacing in option lists
29380 and add line breaks to fix over-long lines.
29382 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
29384 PR middle-end/17660
29386 * extend.texi (Common Variable Attributes): Add xref to GCC
29387 Internals manual to explain mode attribute keywords.
29389 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
29392 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
29393 and Preprocessor Options.
29394 (Options for Linking): Document -pthread here....
29395 (RS/6000 and PowerPC Options): ...not here.
29396 (Solaris 2 Options): ...or here.
29397 * doc/cppopts.texi: Document -pthread.
29399 2017-01-08 Martin Sebor <msebor@redhat.com>
29401 PR middle-end/77708
29402 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
29403 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
29404 New member functions.
29405 (format_directive): Used them.
29407 (pass_sprintf_length::handle_gimple_call): Same.
29408 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
29409 to avoid truncation for any argument.
29410 (extract_affine_mul): Same.
29411 * tree.c (get_file_function_name): Same.
29413 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29415 PR middle-end/77484
29416 * predict.def (PRED_INDIR_CALL): Set to 86.
29418 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
29420 PR preprocessor/54124
29421 * doc/cppopts.texi: Reformat -d subtable to list the full name
29422 of the options. Add cross-reference to the docs for the general
29423 compiler -d options.
29424 * doc/invoke.texi (Developer Options): Add cross-reference to the
29425 preprocessor-specific -d option documentation.
29427 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
29429 PR preprocessor/13498
29430 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
29431 redudant material, and reflect new command-line options.
29432 (System Headers): Likewise.
29434 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
29436 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
29437 -isystem, and -idirafter. Copy-edit.
29438 * doc/cppopts.texi: Copy-edit. Remove contradiction about
29439 default for -ftrack-macro-expansion. Delete obsolete and
29440 badly-formatted implementation details about -fdebug-cpp output.
29441 * doc/cppwarnopts.texi: Copy-edit.
29443 2017-01-07 David Malcolm <dmalcolm@redhat.com>
29446 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
29447 that the transition from a max line width >= 1<<10 to narrower
29448 lines works correctly.
29450 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
29452 * doc/options.texi (PerFunction): New.
29453 * opt-functions.awk (switch_flags): Map both Optimization and
29454 PerFunction to CL_OPTIMIZATION.
29455 * opth-gen.awk: Test for PerFunction flag along with
29457 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
29458 it only when the latter is present. Skip those that don't in
29459 the hash function generator.
29460 * common.opt (fvar-tracking): Mark as PerFunction instead of
29462 (fvar-tracking-assignments): Likewise.
29463 (fvar-tracking-assignments-toggle): Likewise.
29464 (fvar-tracking-uninit): Likewise.
29466 2017-01-07 Jakub Jelinek <jakub@redhat.com>
29468 PR translation/79018
29469 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
29472 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
29475 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
29476 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
29477 TARGET_LONG_JUMP_TABLE_OFFSETS.
29478 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
29479 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
29480 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
29481 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
29482 * config/m68k/m68k.md (tablejump expander): Likewise.
29483 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
29484 TARGET_LONG_JUMP_TABLE_OFFSETS.
29485 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
29486 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
29488 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
29489 David Holsgrove <david.holsgrove@xilinx.com>
29491 * common/config/microblaze/microblaze-common.c
29492 (TARGET_EXCEPT_UNWIND_INFO): Remove.
29493 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
29495 * config/microblaze/microblaze.c (microblaze_must_save_register)
29496 (microblaze_expand_epilogue, microblaze_return_addr): Handle
29498 (microblaze_eh_return): New function.
29499 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
29500 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
29501 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
29502 * config/microblaze/microblaze.md (eh_return): New pattern.
29504 2017-01-06 Jakub Jelinek <jakub@redhat.com>
29506 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
29507 GCC_DIAGNOSTIC_STRINGIFY): Define.
29509 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
29511 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29513 * config/arm/arm.md (<mcrr>): New.
29515 * config/arm/arm.c (arm_arch5te): New.
29516 (arm_option_override): Set arm_arch5te.
29517 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
29519 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
29520 (arm_mcrr_qualifiers): ... this. New.
29521 (MRRC_QUALIFIERS): Define to...
29522 (arm_mrrc_qualifiers): ... this. New.
29523 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
29525 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
29526 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
29527 (MRRCI, mrrc, MRRC): New.
29528 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
29529 VUNSPEC_MRRC2): New.
29531 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29533 * config/arm/arm.md (<mcr>): New.
29535 * config/arm/arm.c (arm_coproc_builtin_available): Add
29536 support for mcr, mrc, mcr2 and mrc2.
29537 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
29538 (arm_mcr_qualifiers): ... this. New.
29539 (MRC_QUALIFIERS): Define to ...
29540 (arm_mrc_qualifiers): ... this. New.
29541 (MCR_QUALIFIERS): Define to ...
29542 (arm_mcr_qualifiers): ... this. New.
29543 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
29545 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
29546 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
29547 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
29548 VUNSPEC_MRC2): New.
29550 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29552 * config/arm/arm.md (*ldc): New.
29556 * config/arm/arm.c (arm_coproc_builtin_available): Add
29557 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
29558 (arm_coproc_ldc_stc_legitimate_address): New.
29559 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
29560 'qualifier_const_pointer'.
29561 (LDC_QUALIFIERS): Define to...
29562 (arm_ldc_qualifiers): ... this. New.
29563 (STC_QUALIFIERS): Define to...
29564 (arm_stc_qualifiers): ... this. New.
29565 * config/arm/arm-protos.h
29566 (arm_coproc_ldc_stc_legitimate_address): New.
29567 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
29568 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
29569 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
29570 stc2, stcl, stc2l): New.
29571 * config/arm/constraints.md (Uz): New.
29572 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
29573 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
29574 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
29575 VUNSPEC_STC2L): New.
29577 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29579 * config/arm/arm.md (<cdp>): New.
29580 * config/arm/arm.c (neon_const_bounds): Rename this ...
29581 (arm_const_bounds): ... this.
29582 (arm_coproc_builtin_available): New.
29583 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
29584 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
29585 (CDP_QUALIFIERS): Define to...
29586 (arm_cdp_qualifiers): ... this. New.
29588 (arm_expand_builtin_args): Add case for 6 arguments.
29589 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
29590 (arm_const_bounds): ... this.
29591 (arm_coproc_builtin_available): New.
29592 * config/arm/arm_acle.h (__arm_cdp): New.
29594 * config/arm/arm_acle_builtins.def (cdp): New.
29596 * config/arm/iterators.md (CDPI,CDP,cdp): New.
29597 * config/arm/neon.md: Rename all 'neon_const_bounds' to
29598 'arm_const_bounds'.
29599 * config/arm/types.md (coproc): New.
29600 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
29601 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
29602 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
29603 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
29605 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29607 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
29608 (UBINOP_QUALIFIERS): New.
29610 (acle_builtin_data): New. Change comment.
29611 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
29612 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
29613 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
29614 arm_acle_builtins.def.
29615 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
29616 (arm_init_acle_builtins): New.
29617 (CRC32_BUILTIN): Remove.
29618 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
29619 crc32cb, crc32ch and crc32cw.
29620 (arm_init_crc32_builtins): Remove.
29621 (arm_init_builtins): Use arm_init_acle_builtins rather
29622 than arm_init_crc32_builtins.
29623 (arm_expand_acle_builtin): New.
29624 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
29625 * config/arm/arm_acle_builtins.def: New.
29627 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29629 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
29630 (arm_builtin_datum): ... this.
29631 (arm_init_neon_builtin): Rename to ...
29632 (arm_init_builtin): ... this. Add a new parameters PREFIX
29633 and USE_SIG_IN_NAME.
29634 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
29635 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
29636 'arm_builtin_datum'.
29637 (arm_init_vfp_builtins): Likewise.
29638 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
29639 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
29640 (arm_expand_neon_args): Rename to ...
29641 (arm_expand_builtin_args): ... this. Rename builtin_arg
29642 enum values and differentiate between ARG_BUILTIN_MEMORY
29643 and ARG_BUILTIN_NEON_MEMORY.
29644 (arm_expand_neon_builtin_1): Rename to ...
29645 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
29646 values, arm_expand_builtin_args and add bool parameter NEON.
29647 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
29648 (arm_expand_vfp_builtin): Likewise.
29649 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
29651 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29653 PR middle-end/77484
29654 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
29655 * predict.c (tree_estimate_probability_bb): Reverse direction of
29656 polymorphic call predictor.
29658 2017-01-06 David Malcolm <dmalcolm@redhat.com>
29660 * passes.c (execute_one_pass): Split out pass-skipping logic into...
29661 (determine_pass_name_match): ...this new function and...
29662 (should_skip_pass_p): ...this new function.
29664 2017-01-06 Nathan Sidwell <nathan@acm.org>
29666 * ipa-visibility.c (function_and_variable_visibility): Reformat
29667 comments and long lines. Remove extrneous if.
29668 * symtab.c (symtab_node::make_decl_local): Fix code format.
29669 (symtab_node::set_section_for_node): Fix comment typo.
29671 2017-01-06 Martin Liska <mliska@suse.cz>
29674 * lra-constraints.c: Rename invariant to lra_invariant.
29675 * predict.c (set_even_probabilities): Initialize e to NULL.
29677 2017-01-05 Martin Sebor <msebor@redhat.com>
29679 PR tree-optimization/78910
29680 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
29681 (format_integer): Correct off-by-one error in the handling
29682 of precision with negative numbers in signed conversions..
29684 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
29686 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
29688 2017-01-05 Jakub Jelinek <jakub@redhat.com>
29690 PR tree-optimization/71016
29691 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
29692 factor_out_conditional_conversion. Formatting fix.
29693 (factor_out_conditional_conversion): Add cond_stmt argument.
29694 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
29695 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
29698 2017-01-05 David Malcolm <dmalcolm@redhat.com>
29700 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
29701 read-rtl-function.o, and selftest-rtl.o.
29702 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
29703 (selftest::aarch64_test_loading_full_dump): New function.
29704 (selftest::aarch64_run_selftests): New function.
29705 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
29706 selftest::aarch64_run_selftests.
29707 * config/i386/i386.c
29708 (selftest::ix86_test_loading_dump_fragment_1): New function.
29709 (selftest::ix86_test_loading_call_insn): New function.
29710 (selftest::ix86_test_loading_full_dump): New function.
29711 (selftest::ix86_test_loading_unspec): New function.
29712 (selftest::ix86_run_selftests): Call the new functions.
29713 * emit-rtl.c (maybe_set_max_label_num): New function.
29714 * emit-rtl.h (maybe_set_max_label_num): New decl.
29715 * function.c (instantiate_decls): Guard call to
29716 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
29717 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
29719 * gensupport.c (gen_reader::gen_reader): Pass "false"
29720 for new "compact" param of rtx_reader.
29721 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
29722 rather than an empty string for NULL strings.
29723 * read-md.c: Potentially include config.h rather than bconfig.h.
29724 Wrap include of errors.h with #ifdef GENERATOR_FILE.
29725 (have_error): New global, copied from errors.c.
29726 (md_reader::read_name): Rename to...
29727 (md_reader::read_name_1): ...this, adding "out_loc" param,
29728 and converting "missing name or number" to returning false, rather
29730 (md_reader::read_name): Reimplement in terms of read_name_1.
29731 (md_reader::read_name_or_nil): New function.
29732 (md_reader::read_string): Handle "(nil)" by returning NULL.
29733 (md_reader::md_reader): Add new param "compact".
29734 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
29735 (md_reader::read_file): New method.
29736 * read-md.h (md_reader::md_reader): Add new param "compact".
29737 (md_reader::read_file): New method.
29738 (md_reader::is_compact): New accessor.
29739 (md_reader::read_name): Convert return type from void to file_location.
29740 (md_reader::read_name_or_nil): New decl.
29741 (md_reader::read_name_1): New decl.
29742 (md_reader::m_compact): New field.
29743 (noop_reader::noop_reader): Pass "false" for new "compact" param
29745 (rtx_reader::rtx_reader): Add new "compact" param.
29746 (rtx_reader::read_rtx_operand): Make virtual and convert return
29747 type from void to rtx.
29748 (rtx_reader::read_until): New decl.
29749 (rtx_reader::handle_any_trailing_information): New virtual function.
29750 (rtx_reader::postprocess): New virtual function.
29751 (rtx_reader::finalize_string): New virtual function.
29752 (rtx_reader::m_in_call_function_usage): New field.
29753 (rtx_reader::m_reuse_rtx_by_id): New field.
29754 * read-rtl-function.c: New file.
29755 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
29756 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
29757 (selftest::verify_three_block_rtl_cfg): New decl.
29758 * read-rtl-function.h: New file.
29759 * read-rtl.c: Potentially include config.h rather than bconfig.h.
29760 For host, include function.h, memmodel.h, and emit-rtl.h.
29761 (one_time_initialization): New function.
29762 (struct compact_insn_name): New struct.
29763 (compact_insn_names): New array.
29764 (find_code): Handle insn codes in compact dumps.
29765 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
29766 (bind_subst_iter_and_attr): Likewise.
29767 (add_condition_to_string): Likewise.
29768 (add_condition_to_rtx): Likewise.
29769 (apply_attribute_uses): Likewise.
29770 (add_current_iterators): Likewise.
29771 (apply_iterators): Likewise.
29772 (initialize_iterators): Guard usage of apply_subst_iterator with
29773 #ifdef GENERATOR_FILE.
29774 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
29775 (md_reader::read_mapping): Likewise.
29776 (add_define_attr_for_define_subst): Likewise.
29777 (add_define_subst_attr): Likewise.
29778 (read_subst_mapping): Likewise.
29779 (check_code_iterator): Likewise.
29780 (rtx_reader::read_rtx): Likewise. Move one-time initialization
29782 (one_time_initialization): New function.
29783 (rtx_reader::read_until): New method.
29784 (read_flags): New function.
29785 (parse_reg_note_name): New function.
29786 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
29787 Handle reuse_rtx ids.
29788 Wrap iterator lookup within #ifdef GENERATOR_FILE.
29789 Add parsing support for RTL dumps, mirroring the special-cases in
29790 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
29791 values, and calling handle_any_trailing_information.
29792 (rtx_reader::read_rtx_operand): Convert return type from void
29793 to rtx, returning return_rtx. Handle case 'e'. Call
29794 finalize_string on XSTR and XTMPL fields.
29795 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
29796 "(nil)" values were omitted. Call the postprocess vfunc on the
29798 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
29799 class ctor. Initialize m_in_call_function_usage. Call
29800 one_time_initialization.
29801 * rtl-tests.c (selftest::test_uncond_jump): Call
29802 set_new_first_and_last_insn.
29803 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
29804 * selftest-rtl.c: New file.
29805 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
29806 (selftest::get_insn_by_uid): New decl.
29807 * selftest-run-tests.c (selftest::run_tests): Call
29808 read_rtl_function_c_tests.
29809 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
29810 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
29813 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
29815 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
29816 operands in a special way. Assert that pos+len <= mode precision.
29818 2017-01-05 Jakub Jelinek <jakub@redhat.com>
29820 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
29821 3 argument Alias with unlimited for the negative form.
29822 (fno-vect-cost-model): Removed.
29824 2017-01-05 Martin Liska <mliska@suse.cz>
29826 * hsa-gen.c (gen_hsa_divmod): New function.
29827 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
29829 2017-01-05 Martin Liska <mliska@suse.cz>
29832 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
29835 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29837 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
29838 small constant length operands.
29840 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29842 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
29843 between loop iterations.
29845 2017-01-05 Martin Liska <mliska@suse.cz>
29848 * gimplify.c (gimplify_decl_expr): Compare to
29849 asan_poisoned_variables instread of checking flags.
29850 (gimplify_target_expr): Likewise.
29851 (gimplify_expr): Likewise.
29852 (gimplify_function_tree): Conditionally initialize
29853 asan_poisoned_variables.
29855 2017-01-04 Jeff Law <law@redhat.com>
29857 PR tree-optimizatin/78812
29858 * rtl.h (contains_mem_rtx_p): Prototype.
29859 * ifcvt.c (containts_mem_rtx_p): Move from here to...
29860 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
29861 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
29862 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
29863 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
29865 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29867 * input.c (assert_char_at_range): Default-initialize actual_range.
29869 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29871 * df-scan.c (df_ref_create_structure): Make regno unsigned,
29872 to match the caller.
29874 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29876 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
29877 insns after final jump in test to emit dummy move.
29879 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29881 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
29882 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
29884 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29886 * multiple_target.c (create_dispatcher_calls): Init e_next.
29887 * tree-ssa-loop-split.c (split_loop): Init border.
29888 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
29891 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
29896 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
29897 (altivec_register_operand): Do not return true if the operand
29898 contains a SUBREG mixing SImode and SFmode.
29899 (vsx_register_operand): Likewise.
29900 (vsx_reg_sfsubreg_ok): New predicate.
29901 (vfloat_operand): Do not return true if the operand contains a
29902 SUBREG mixing SImode and SFmode.
29903 (vint_operand): Likewise.
29904 (vlogical_operand): Likewise.
29905 (gpc_reg_operand): Likewise.
29906 (int_reg_operand): Likewise.
29907 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
29908 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
29909 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
29911 (rs6000_emit_move_si_sf_subreg): New helper function.
29912 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
29913 fixup SUBREGs involving SImode and SFmode.
29914 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
29915 numbers for the new peephole2 optimization.
29916 (peephole2 for SFmode unions): New peephole2 to optimize cases in
29917 the GLIBC math library that do AND/IOR/XOR operations on single
29918 precision floating point.
29919 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
29920 target macros to say whether we need to avoid SUBREGs mixing
29922 (TARGET_ALLOW_SF_SUBREG): Likewise.
29923 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
29924 (UNSPEC_SI_FROM_SF): Likewise.
29925 (iorxor): Change spacing.
29926 (and_ior_xor): New iterator for AND, IOR, and XOR.
29927 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
29928 (movdi_from_sf_zero_ext): Likewise.
29929 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
29930 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
29931 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
29932 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
29933 (fms<mode>4): Likewise.
29934 (fnma<mode>4): Likewise.
29935 (fnms<mode>4): Likewise.
29936 (nfma<mode>4): Likewise.
29937 (nfms<mode>4): Likewise.
29939 2017-01-04 Marek Polacek <polacek@redhat.com>
29942 * doc/invoke.texi: Document -Wpointer-compare.
29944 2017-01-04 Jakub Jelinek <jakub@redhat.com>
29946 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
29949 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
29950 descriptions for -gdwarf-5 and emit them as uleb128 instead of
29953 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
29956 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
29957 documentation of the powerpc_popcntb_ok attribute.
29958 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29959 code to issue warning messages if a requested CPU configuration is
29960 not supported by the binary (assembler and loader) toolchain.
29961 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
29962 made to define a built-in function that has been disabled.
29963 (paired_init_builtins): Add assertion to prevent ICE if attempt is
29964 made to define a built-in function that has been disabled.
29965 (altivec_init_builtins): Add comment explaining why definition
29966 of the DST built-in functions is not preceded by an assertion
29967 check. Add assertions to prevent ICE if attempts are made to
29968 define an altivec predicate or an abs* built-in function that has
29970 (htm_init_builtins): Add comment explaining why definition of the
29971 htm built-in functions is not preceded by an assertion check.
29973 2017-01-04 Jeff Law <law@redhat.com>
29975 PR tree-optimizatin/67955
29976 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
29977 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
29978 the points-to solution does not include pt_null. Use DECL_PT_UID
29981 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
29983 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
29984 Use gen_int_mode instead of gen_lopwart for const_int operands.
29986 2017-01-04 Jakub Jelinek <jakub@redhat.com>
29988 PR tree-optimization/71563
29989 * match.pd: Simplify X << Y into X if Y is known to be 0 or
29990 out of range value - has low bits known to be zero.
29992 2017-01-04 Alan Modra <amodra@gmail.com>
29994 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
29995 * configure: Regenerate.
29996 * config.in: Regenerate.
29998 2017-01-04 Jakub Jelinek <jakub@redhat.com>
30001 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
30002 a substring of the message, but strcmp with the whole message. Ifdef
30003 ENABLE_NLS, translate the message first using dgettext.
30005 2017-01-03 Jeff Law <law@redhat.com>
30007 PR tree-optimizatin/78856
30008 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
30009 (mark_threaded_blocks): Remove code to truncate thread paths that
30010 cross multiple loop headers. Instead invalidate the cached loop
30011 iteration information and handle case of a thread path walking
30012 into an irreducible region.
30014 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
30017 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
30018 assertions. Add support for doing the signbit if the IEEE 128-bit
30019 floating point value is in a GPR.
30020 * config/rs6000/rs6000.md (Fsignbit): Delete.
30021 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
30022 Update the length attribute if the value is in a GPR.
30023 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
30024 the sign or zero extension instruction, since the value is always 0/1.
30025 (signbit<mode>2_dm2): Delete using <Fsignbit>.
30028 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
30029 extracting SImode to a GPR register so that we can generate a
30030 store, limit the vector to be in a traditional Altivec register
30031 for the vextuwrx instruction.
30033 2017-01-03 Ian Lance Taylor <iant@google.com>
30035 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
30037 2017-01-03 Martin Sebor <msebor@redhat.com>
30039 PR tree-optimization/78696
30040 * gimple-ssa-sprintf.c (format_floating): Correct handling of
30041 precision. Use MPFR for %f for greater fidelity. Correct handling
30043 (pass_sprintf_length::compute_format_length): Set width and precision
30044 specified by asrerisk to void_node for vararg functions.
30045 (try_substitute_return_value): Adjust dump output.
30047 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
30049 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
30051 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
30053 * doc/invoke.texi (SPARC options): Document -mlra as the default.
30054 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
30055 -mlra/-mno-lra was passed to the compiler.
30057 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
30059 PR rtl-optimization/65618
30060 * emit-rtl.c (try_split): Move initialization of "before" and
30061 "after" to just before the call to emit_insn_after_setloc.
30063 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
30065 * doc/md.texi (Standard Names): Remove reference to Java frontend.
30067 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
30069 * dwarf2out.c (gen_enumeration_type_die): When
30070 -gno-strict-dwarf, add a DW_AT_encoding attribute.
30072 2017-01-03 Jakub Jelinek <jakub@redhat.com>
30074 PR tree-optimization/78965
30075 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
30076 Change first argument from const call_info & to call_info &. For %n
30077 set info.nowrite to false.
30079 PR middle-end/78901
30080 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
30081 possibly throwing calls.
30083 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
30084 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
30085 and fns handling, rather than in a separate case SSA_NAME.
30087 2017-01-02 Jeff Law <law@redhat.com>
30089 * config/darwin-driver.c (darwin_driver_init): Const-correctness
30090 fixes for first_period and second_period variables.
30092 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
30095 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
30096 (*insvqi_1): New insn pattern.
30097 (*insvqi_1_mem_rex64): Ditto.
30098 (*insvqi_2): Ditto.
30099 (*insvqi_3): Rename from *insvqi.
30101 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
30103 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
30105 * doc/cfg.texi (Edges): Remove reference to Java.
30106 (Maintaining the CFG): Ditto.
30108 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30110 PR middle-end/77674
30111 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
30112 transparent aliases.
30114 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30116 PR middle-end/77484
30117 * predict.def (PRED_CALL): Update hitrate.
30118 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
30119 * predict.c (tree_estimate_probability_bb): Split CALL predictor
30120 into direct/indirect/polymorphic variants.
30122 2017-01-01 Jakub Jelinek <jakub@redhat.com>
30124 Update copyright years.
30126 * gcc.c (process_command): Update copyright notice dates.
30127 * gcov-dump.c (print_version): Ditto.
30128 * gcov.c (print_version): Ditto.
30129 * gcov-tool.c (print_version): Ditto.
30130 * gengtype.c (create_file): Ditto.
30131 * doc/cpp.texi: Bump @copying's copyright year.
30132 * doc/cppinternals.texi: Ditto.
30133 * doc/gcc.texi: Ditto.
30134 * doc/gccint.texi: Ditto.
30135 * doc/gcov.texi: Ditto.
30136 * doc/install.texi: Ditto.
30137 * doc/invoke.texi: Ditto.
30139 Copyright (C) 2017 Free Software Foundation, Inc.
30141 Copying and distribution of this file, with or without modification,
30142 are permitted in any medium without royalty provided the copyright
30143 notice and this notice are preserved.